MTC Oncology

9781842143094-FM 8/8/07 12:09 PM Page i
Multidetector Computed
Tomography in Oncology
CT Perfusion Imaging
9781842143094-FM 8/8/07 12:09 PM Page ii
9781842143094-FM 8/8/07 12:09 PM Page iii
Multidetector Computed
Tomography in
Oncology
CT Perfusion Imaging
Editors
Kenneth A Miles MBBS FRCR MSc MD FCRP

Brighton and Sussex Medical School
Brighton
UK
Charles-André Cuenod MD PhD
Laboratoire de Recherche en Imagerie
Université Paris V
Paris
France
Foreword by Dame Janet Husband
9781842143094-FM 8/8/07 12:09 PM Page iv
© 2007 Informa UK Ltd
First published in the United Kingdom in 2007 by Informa Healthcare, Telephone

House, 69–77 Paul Street, London EC2A 4LQ. Informa Healthcare is a trading division
of Informa UK Ltd. Registered Office: 37/41 Mortimer Street, London W1T 3JH.
Registered in England and Wales number 1072954.
Tel: +44 (0)20 7017 5000

Fax: +44 (0)20 7017 6699
Website: www.informahealthcare.com
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise, without the prior permission of the
publisher or in accordance with the provisions of the Copyright, Designs and Patents
Act 1988 or under the terms of any licence permitting limited copying issued by the
Copyright Licensing Agency, 90 Tottenham Court Road, London W1P 0LP.
Although every effort has been made to ensure that all owners of copyright material
have been acknowledged in this publication, we would be glad to acknowledge in
subsequent reprints or editions any omissions brought to our attention.
Although every effort has been made to ensure that drug doses and other information
are presented accurately in this publication, the ultimate responsibility rests with the
prescribing physician. Neither the publishers nor the authors can be held responsible
for errors or for any consequences arising from the use of information contained herein.
For detailed prescribing information or instructions on the use of any product or
procedure discussed herein, please consult the prescribing information or instructional
material issued by the manufacturer.
A CIP record for this book is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Data available on application
ISBN-10: 1-84214-309-3
ISBN-13: 978-1-84214-309-4
Distributed in North and South America by

Taylor & Francis
6000 Broken Sound Parkway, NW, (Suite 300)
Boca Raton, FL 33487, USA
Within Continental USA

Tel: 1 (800) 272 7737; Fax: 1 (800) 374 3401
Outside Continental USA
Tel: (561) 994 0555; Fax: (561) 361 6018
Email: orders@crcpress.com
Distributed in the rest of the world by

Thomson Publishing Services
Cheriton House
North Way
Andover, Hampshire SP10 5BE, UK
Tel: +44 (0)1264 332424
Email: tps.tandfsalesorder@thomson.com
Composition by Cepha Imaging Pvt Ltd, Bangalore, India.

Printed and bound in India by Replika Press Pvt Ltd
9781842143094-FM 8/8/07 12:09 PM Page v
To Elizabeth, Matthew, Benjamin, and Samuel …

9781842143094-FM 8/8/07 12:09 PM Page vi
9781842143094-FM 8/8/07 12:09 PM Page vii
Contents
List of contributors ix
Foreword xiii
Preface xv
1 Computed tomography perfusion: a historical perspective 1

Anne E Miles and Kenneth A Miles
2 Scientific basis and validation 15

Ting-Yim Lee and Errol Stewart
3 Image acquisition and contrast enhancement 47

protocols for perfusion CT
Kenneth A Miles
4 Image processing 61
Ting-Yim Lee, Xiaogang Chen, and Kenneth A Miles
5 Angiogenesis, tumor perfusion, and cancer management 73

William W Li, Aliya Jiwani, and Michelle Hutnik
6 Tumors of the brain, head, and neck 89

Part A: Perfusion CT imaging in cerebral neoplasms
Karim Samji and Kenneth A Miles
Part B: Perfusion CT in head and neck cancer
Robert Hermans
7 Perfusion CT applications in lung cancer 111

Kwun M Fong, Rayleen V Bowman, Ian A Yang,
and Kenneth A Miles
9781842143094-FM 8/8/07 12:09 PM Page viii
viii Contents
8 Tumors of the gastro-intestinal tract 129

Part A: Rectal cancer
Massimo Bellomi and Giuseppe Petralia
Part B: Other gastrointestinal tumors
Kenneth A Miles
9 Tumors of the urogenital tract 147

Elizabeth P Ives and Ethan J Halpern
10 CT perfusion of lympth nodes 163

Kenneth A Miles
11 CT perfusion of liver metastases and early 173

detection of micrometastases
Charles-André Cuenod, Laure Fournier,
Daniel Balvay, and Kenneth A Miles
12 Beyond RECIST: perfusion CT in evaluating 197

treatment response and complications
Natalie Charnley and Kenneth A Miles
13 Perfusion CT–PET: opportunities for combined 215

assessment of tumor vascularity and metabolism
Kenneth A Miles
9781842143094-FM 8/8/07 12:09 PM Page ix
Contributors
Daniel Balvay PhD Xiaogang Chen PhD

Laboratoire de Recherche Radiology and Nuclear Medicine
en Imagerie Department
Université Paris V The University of Western
Paris Ontario
France London, ON
Canada
Massimo Bellomi MD
School of Medicine Charles-André Cuenod MD PhD
University of Milan Laboratoire de Recherche en
and Imagerie
Department of Radiology Université Paris V
European Institute of Oncology Paris
Milan France
Italy
Kwun M Fong MBBS FRACP PhD
Rayleen V Bowman MBBS PhD Department of Thoracic Medicine
FRACP The Prince Charles Hospital
The University of Queenland Brisbane
Brisbane Australia
Australia
Laure Fournier MD PhD
Natalie Charnley MBChB MRCP Laboratoire de Recherche
FRCR en Imagerie
The University of Manchester Université Paris V
Wolfson Molecular Imaging Paris
Centre France
Manchester
UK
9781842143094-FM 8/8/07 12:09 PM Page x
x Contributors
Ethan J Halpern MD William W Li

Department of Radiology The Angiogenesis Foundation
Thomas Jefferson University Cambridge, MA
Jefferson Medical College USA
Philadelphia, PA
USA Anne E Miles BSc PGCE MA
University of London
Robert Hermans MD PhD London
Department of Radiology UK
University Hospitals Leuven
Leuven Kenneth A Miles MBBS FRCR MSc
Belgium MD FCRP
Brighton and Sussex Medical
Michelle Hutnik School
The Angiogenesis Foundation Brighton
Cambridge, MA UK
USA
Blake Murphy BSC
Elizabeth P Ives MD Radiology and Nuclear Medicine
Department of Radiology Department
Thomas Jefferson University The University of Western
Philadelphia, PA Ontario
USA London, ON
Canada
Aliya Jiwani
The Angiogenesis Foundation Giuseppe Petralia MD
Cambridge, MA Department of Radiology
USA European Institute of Oncology
Milan
Ting-Yim Lee PhD FCCPM Italy
Radiology and Nuclear Medicine
Department Karim Samji MBBS BSc
The University of Western Brighton and Sussex University
Ontario Hospitals NHS Trust
London, ON Brighton
Canada UK
9781842143094-FM 8/8/07 12:09 PM Page xi
Contributors xi
Errol Stewart BSC Ian A Yang MBBS PhD FRACP

Radiology and Nuclear Medicine Department of Thoracic
Department Medicine
The University of Western The Prince Charles Hospital
Ontario Brisbane
London, ON Australia
Canada
9781842143094-FM 8/8/07 12:09 PM Page xii
9781842143094-FM 8/8/07 12:09 PM Page xiii
Foreword
The extraordinary advances in medical technology over recent years

have placed imaging at the center of cancer diagnosis and assessment
but today the information used to direct patient management is based
almost entirely on morphological assessment. However, we are now at
the brink of a new and exciting era in which functional data will
become an integral component of routine tumor imaging.
It is well recognized that a tumor cannot grow beyond the size of
1mm3 without a blood supply and that assessment of tumor vasculature
provides a measure of tumour aggressiveness as well as insight into
other factors related to prognosis, prediction of response to treatment
and risk of recurrence.
This text brings imaging of tumor vasculature into the domain of
leading edge clinical practice and describes the added value and limita-
tions of current perfusion techniques in individual tumor types. While
Magnetic Resonance Imaging (MRI) provides a unique tool for assess-
ing tumors using a multifunctional approach, multidetector CT (MDCT)
is more widely available for staging tumors and indeed remains the
workhorse of cancer imaging today. It is appropriate therefore that
MDCT should be exploited to provide both morphological and func-
tional measurements of tumor vasculature in the routine assessment of
patients with cancer.
This approach is a welcome step forward in striving to reach the goal
of providing a detailed portrait of the morphological and functional
aspects of a tumor prior to therapy. Opportunities for developing
multifunctional assessment of tumors by a combination of PET and CT
data will take us further down this road, particularly as we look forward
to the introduction of new tracers which will allow interrogation of
multiple biological processes.
9781842143094-FM 8/8/07 12:09 PM Page xiv
xiv Foreword
Professor Miles has undertaken pioneering work in the development

of perfusion imaging with CT and here he brings together the views of
a team of highly regarded world experts. Together they present a con-
temporary analysis of the current evidence of the role of perfusion
MDCT in cancer medicine thus providing the foundation on which to
build a robust framework for the future.
It is hoped that in the not too distant future these techniques will be
applied to patients with cancer routinely and that this will lead to
improved survival and better patient outcomes.
Professor Dame Janet Husband DBE FMedSci PRCR FRCP

The Institute of Cancer Research
Royal Marsden NHS Trust
9781842143094-FM 8/8/07 12:09 PM Page xv
Preface
It is now more than 25 years since Leon Axel proposed a method for
determination of cerebral blood flow from rapid-sequence contrast-
enhanced computed tomography (CT). Today, the availability of rapid
imaging with multidetector CT systems and commercial analysis soft-
ware has made perfusion imaging with CT an everyday technique for
clinical practice.
CT remains an essential tool in the assessment of patients with
cancer, not only for diagnosis but also for assessment of disease extent
and severity, and for the evaluation of response to treatment. Perfusion
CT is readily performed as an adjunct to conventional CT, providing
valuable information about tumor vascularity. In many ways, perfusion
CT is not unlike CT angiography, but depicts the functional status of the
tumor circulation at tissue level rather than visualizing the morphology
of discrete vessels. By reflecting the processes of tumor angiogenesis,
this additional information can aid in diagnosis, assess tumor aggres-
sion, and help to overcome some of the limitations associated with
morphological criteria for evaluation of tumor response.
The development of perfusion CT also links with another recent
advance in cancer imaging, the introduction of integrated positron
emission tomography (PET)–CT systems. The increasing use of intra-
venous contrast material during PET–CT can be extended to include a
CT perfusion study. In this way, it is now possible to depict tumor mor-
phology, perfusion, and glucose metabolism to provide an exceptionally
detailed assessment of tumor biology in a single examination.
This book is the first to be dedicated solely to the application of
perfusion CT in oncology. The aim is to provide the technical knowledge
required to reliably obtain CT perfusion images of tumors, to give an
understanding of the pathophysiology of tumor angiogenesis and its
9781842143094-FM 8/8/07 12:09 PM Page xvi
xvi Preface
relationship to contrast enhancement on CT, and to outline the ability

of perfusion CT to enhance diagnosis, prognosis, and therapy monitoring
for patients with cancer. The technique is also portrayed in its historical
context. The book will therefore be of interest to radiologists and radi-
ographers currently using perfusion CT or considering its introduction
to their institution. The information will also be valuable to clinicians
treating patients with cancer and to researchers involved in the devel-
opment of new cancer therapies.
Ken Miles MD
Brighton and Sussex Medical School
9781842143094-Ch01 8/8/07 4:26 PM Page 1
1
Perfusion computed
tomography: a historical
perspective
Anne E Miles and Kenneth A Miles
Perfusion computed tomography (CT) has been made possible by the

joining together of two great medical specialities: anatomy and physi-
ology. It is important for us today as doctors and scientists to look back
into history and to our predecessors, at the state of knowledge from
ancient up to modern times, as well as looking forward to the possibil-
ities and dreams to which we aspire. J B Thornton wrote: ‘The more we
treat the theories of our predecessors as myths, the more inclined we
shall be to treat our own theories as dogmas.1 By looking back through
history, we are able to see the grand schemes and ideas developing
over time, and the continuity of ideas from great thinkers which can be
added to and extended to attain new levels of knowledge by our own
thinkers today. This chapter does not attempt to mention all the impor-
tant thinkers and achievers who have helped anatomy and physiology
become what they are together. It would be an impossible task. Rather,
by dipping in and out of history, we will attempt to cover some of
the more major scientists, philosophers, and physicians who have
contributed to making CT perfusion what it is today.
ANATOMICAL AND PHYSIOLOGICAL KNOWLEDGE

IN ANCIENT CIVILIZATIONS
Many of the medical accomplishments of the great civilizations of

antiquity have been sadly ignored in the West because of the lack of
accurate records, and problems with deciphering material that has been,
and still is being, discovered. One of the earliest Egyptian papyruses
9781842143094-Ch01 8/8/07 4:26 PM Page 2
2 Multidetector computed tomography in oncology
(Edwin Smith Papyrus) is probably a copy of one that was first written
between 3000 and 2500 BC. Translated in 1930, this papyrus is essen-
tially a surgical document, and, for the first time in recorded history, the
word ‘brain’ is mentioned. The heart is also mentioned as the center of
a distributing system of vessels which pulsates.2 A papyrus purchased
by George Embers at Thebes in Egypt, not a copy but an original
document written about 16 centuries before the Christian Era, is now
considered to be the oldest known anatomical document. It mentions
that ‘there are vessels from it [the heart] to all the members’.2 A more
accurate impression of the structure of the human body was obtained
in ancient Egypt during the period of the New Kingdom (late dynasty
XVII through dynasty XX). This was probably possible because of the
practice of embalming and mummification which had reached its
highest level at that time.
In China, one of the oldest civilizations known, the doctrine of
Confucianism imposed restrictions upon dissection similar to those later
experienced by Galen. Dissection was not practiced, in order not to
defile the human body. Nevertheless, the medical scholars of ancient
China revealed through their writings a keen sense of awareness of the
human body for the treatment of disease. Huang Ti (2600 BC) is the
father of Chinese medicine. In his ‘Canon of Medicine’ or ‘Nei Ch’ing’
he writes that ‘all the blood of the body is under control of the heart.
The heart is in accord with the pulse. The pulse regulates all the blood
and the blood current flows in a continuous circle and never stops’.2
Remarkably, this document clearly recognizes a relationship between
blood, pulse, and the heart. It took William Harvey in the 17th century
to confirm this knowledge to the Western world.
India was yet another ancient civilization in which traditional heal-
ing methods and practical skills were remarkably advanced. Much of
this knowledge spread slowly via Asia, and reached Europe during the
Middle Ages as a result of translations that were made by Persian and
Arab scholars in the 11th century.
In the Western world, Hippocrates (about 460–377 BC), born on the
island of Cos in the Aegean, is considered by many to be the greatest
of all physicians and ‘the Father of Medicine’. However, much of the
‘Hippocratic Corpus’, a large collection of philosophical, scientific, and
medical works, was written between 300 and 200 BC by a collection
of physicians, probably of the medical school of Cos, rather than
Hippocrates himself. These writings were further compiled and edited
by the scholars of the library at Alexandria.2 Hippocrates is believed
to have disliked dissection, and his descriptions were probably based
on visual examinations of the body surface and the investigation
9781842143094-Ch01 8/8/07 4:26 PM Page 3
Perfusion computed tomography: a historical perspective 3
of wounds. His knowledge of internal organs was thus largely specula-

tive. Indeed, in the ancient Greek world, science was considered to
be mainly a metaphysical field of study, with direct observations of
phenomena being of secondary importance.1
Aristotle (384–322 BC) was the greatest natural philosopher of his
era. Although not a physician, he still contributed much to the study of
medicine. Born in the city of Stagyra, and son of a court physician to
king Philip of Macedonia, Aristotle carried out extensive and fairly accu-
rate studies, including dissections, on a wide range of animals. Like
Hippocrates, Aristotle’s knowledge of the human body was derived
from external observations and speculation based on his animal dissec-
tions. Aristotle mentioned the aorta for the first time in history.2
He stated that it arose from the heart and not from the head and brain
as was previously stated by Polybus, Hippocrates’ son-in-law. He also
believed, however, that the mind was held in the heart. Charles Darwin
thought that Aristotle was the world’s greatest natural scientist, and that
Aristotle laid the foundation for comparative anatomy as a result of the
animal dissections he carried out and his speculations about the layout
of the human body.
THE DEVELOPMENT OF MODERN ANATOMY
Anatomy is one of the oldest branches of medicine, and in Western

civilizations, the physician and philosopher Claudius Galen is often the
most celebrated anatomist of antiquity. Galen was born at Pergamum in
AD 129. His father, Aelius Nicon, an architect and builder with an inter-
est in mathematics, logic, and astronomy, planned for his son to study
philosophy or politics, the traditional pursuits of the cultured governing
clan into which he had been born. But the healing god Asclepius
apparently intervened in one of Nicon’s dreams. He was to allow Galen
to study medicine.
Galen studied medicine for a total of 12 years in Smyrna, Corinth,
and at Alexandria. When he returned to Pergamum in AD 157, he
worked as a physician in a gladiator school for 3 or 4 years: a very
prestigious appointment. These few years provided him with valuable
practical experience in trauma and sports medicine. He later regarded
wounds as ‘windows into the body’, as dissection was forbidden in
imperial Rome. From AD 162, Galen lived mainly in Rome, where he
gained a reputation as an experienced physician, and eventually
became a court physician to emperor Marcus Aurelius. The rest of
his life was spent in the royal court, writing and experimenting.
9781842143094-Ch01 8/8/07 4:26 PM Page 4
Galen expanded his knowledge partly by experimenting with live ani-

mals. A ‘party piece’ he frequently exhibited involved publicly dissect-
ing a living pig by cutting its nerve bundles one at a time. Eventually he
would cut a laryngeal nerve (now also known as Galen’s nerve) and the
pig stopped squealing. It is interesting that acts considered as unaccept-
ably cruel today were considered necessary or even entertaining then.
Galen transmitted Hippocratic medicine all the way to the
Renaissance, describing the philosopher’s system of four bodily humors
linked to the four classical elements. Galen’s anatomical writings were
riddled with errors and shortcomings, but he made some important
findings which should not be overlooked. He demonstrated that arter-
ies carry blood not air, and he made the first Western studies about
nerve functions, brain and heart. He also argued that the mind was in
the brain, not in the heart, as Aristotle had claimed. Galen also believed
that a tumor might form where there was too much blood in the veins,3
perhaps heralding the later concept of angiogenesis. However, he did
not recognize blood circulation, and thought that venous and arterial
systems were separate. Despite such errors, Galen’s anatomical writings
remained unchallenged up to the time of Versalius in the 16th century.
His works took on an almost Christian sacredness. To criticize any of
them was life-endangering heresy.4 The Royal College of Physicians
of London in 1559 even made one of its members, Dr John Geynes,
retract his statement that there were 22 inaccurate passages in the
works of Galen.
The study of anatomy seemed to die between the fall of Rome and
the Renaissance. There appear to be several possible reasons for this
decline. First, with Galen’s authority dominating medicine, scientists no
longer bothered to experiment, and studies into anatomy and physiol-
ogy stopped. The dying of the Roman Empire also appeared to deaden
intellectual, artistic, and scientific activity, and the almost universal pro-
hibition of dissection of human-beings caused problems. Although the
art of healing was highly regarded in the Middle Ages, anatomy, being
concerned with the dead, was considered immoral and irreligious.
It was not until the start of the Renaissance that a few Italian city-states
(Bologna, Padua, and Pavia) began to permit the dissection of a few exe-
cuted criminals each year. One of the first of these human dissections was
carried out at the University of Bologna by the anatomist Mondino de
Luzzi (1276–1326). The subject was an executed female, and Mondino sat
reading from the work of Galen whilst his assistant performed the dissec-
tion. Mondino’s ‘Anathomia’ was the first modern work to deal exclu-
sively with anatomy. However, he never questioned the authoritative
writings of Galen, even when the findings were contradictory.
9781842143094-Ch01 8/8/07 4:26 PM Page 5
Many famous artists of the early Renaissance pursued the study of

the human body, including actual dissections, in order to show the
beauty of the human form in an accurate and realistic manner. This
union of art and science brought the study of human anatomy onto a
new, promising, and irreversible course2. In more than 750 anatomical
drawings of the musculoskeletal, vascular, nervous, and urogenital sys-
tems, Leonardo da Vinci produced work of unchallenged artistic beauty
and scientific accuracy that quickly transcended the needs of the artist
and drifted into the scientific pursuit of anatomy for its own end.
Leonardo placed a great deal of importance on the laws of geometry
and mechanics as applied to the human form. He remarked: ‘let no man
read me who is not a mathematician. No human investigation can lay
claim to being true science unless it can stand the test of mathematical
demonstration. The man who undervalues mathematics nourishes
himself upon confusion’. The mathematician and physiologist Fick,
whom we will discuss later, would have been delighted at Leonardo’s
sentiments.
Andreas Vesalius (1514–1564) has been described as the most com-
manding figure in European medicine between Galen and Harvey.2
Born in Brussels, Belgium, he came from a distinguished family of
physicians. After studying at Pedagogium Castri and Collegium
Trilingue at Louvain, he entered the distinguished but extremely
conservative medical school of the University in Paris in 1536. Few dis-
sections were carried out at the University, as all teaching continued to
be based on Galenism. In 1537, Versalius returned to Louvain from
Paris, and the following year he conducted one of the first human dis-
sections to be held in the city for 18 years. He later traveled to Venice
and received his Doctor of Medicine degree from the University of
Padua. On the following day, the senate of Venice appointed Versalius
Professor of Surgery, with the responsibility also for Anatomy at the
University. This progression to professor has to be rapid by anyone’s
standard!
Although a Galenist at first, by studying the human body itself,
Versalius began to reveal differences from the descriptions made by
Galen. In 1540, Versalius made a dramatic demonstration in Bologna of
the skeletons of a man and an ape and demonstrated more than 200
differences where Galen was mistaken with respect to the human body
but not to that of the ape. Versalius’ book ‘Fabrica’ (finished in 1542)
was based upon actual dissection and original observations. Although
his work was initially considered outrageous by the Galenists of the
time, it was rapidly acknowledged as an outstanding exposition of
the true structure of the human body. Its publication marked a new era
9781842143094-Ch01 8/8/07 4:26 PM Page 6
in medicine and the beginning of modern anatomy. Versalius died in

1564 upon return from a hazardous pilgrimage to Jerusalem, and after
the publication of ‘Fabrica’ no other noteworthy anatomical book was
to appear for over a century.
ANATOMY AND IMAGING
The ability to depict the internal anatomy of the human body through
non-invasive imaging rather than dissection or surgery dates back to the
discovery of X-rays in 1895 by Wilhelm Conrad Röntgen (1845–1923).4
Whilst Chair of Physics at the University of Würzburg, Röntgen had
been studying the phenomena associated with the passage of electric-
ity through a gas at extremely low pressure, using an evacuated glass
tube developed by Sir William Crookes (1832–1919). On November 8,
1895, Röntgen had enclosed the Crookes tube in a sealed, thick black
carton to exclude all light. When the electric current was switched on,
he noticed that a paper plate coated with barium platinocyanide began
to fluoresce, even when it was as far as 2 meters away from the tube.
He deduced the existence of hitherto unknown rays, dubbing them
‘X-rays’. The now famous radiograph of the hand of Röntgen’s wife,
Bertha, was taken within 1 month of his discovery. The shadow cast by
her bones and the ring on her finger were clearly visible, surrounded
by a penumbra produced by the flesh. In December of that year,
Röntgen published his findings in the Proceedings of the Würzburg
Physical–Medical Society in an article entitled ‘On a New Kind of Ray:
A Preliminary Communication’. By January 1896, X-rays were being
used in several countries around the world to diagnose fractures and to
detect radio-opaque foreign bodies such as bullets. The use of X-rays
for diagnosis and therapy grew rapidly thereafter. In 1901, Röntgen was
awarded the Nobel Prize for Physics in recognition of his discovery.
The means of using X-rays to depict the anatomy of the circulation
came in the 1920s with the development of cerebral angiography
by the Portuguese neurosurgeon Egas Moniz (1874–1955).5 Whilst
Professor of Neurology at Lisbon, Moniz had sought to identify a radio-
opaque dye that was non-toxic and would pass through the capillaries
without causing a blockage. Working first on animals and cadavers
before moving to human subjects, Moniz tried injections of air, bro-
mides, and iodides. The X-ray attenuating properties of the iodine atom
form the basis of contrast agents used for angiography and CT perfusion
today (Figure 1.1). However, iodine-containing compounds with lower
toxicity than the simple iodides used by Moniz were developed
9781842143094-Ch01 8/8/07 4:26 PM Page 7
Figure 1.1 Venous phase of a cerebral angiogram depicting the great cerebral
vein of Galen. Although Galen had a detailed knowledge of anatomy, he
believed that the arterial and venous systems were separate. The ability to depict
the anatomy of the cerebral circulation non-invasively came with the development
of cerebral angiography by Moniz in the 1920s
in the 1930s, when it was noticed that iodine-containing products

intended to improve the treatment of syphilis, were radio-opaque when
excreted in the urinary system. Based on the results of his technique,
Moniz also described the formation of new vessels in brain tumors.6
In 1949 Moniz received the Nobel Prize for Physiology or Medicine,
more for his discovery of prefrontal leukotomy than for the development
of cerebral angiography.
The development of CT by Hounsfield and Ambrose in 1973 repre-
sented a major advance in the ability of imaging to demonstrate
anatomy.7,8 By providing a means to depict the human body in cross-
section, CT was able to reveal brain structures that had been invisible
on radiographs until then. At the time of his discovery, Sir Godfrey
Hounsfield (1919–2004) was an electrical engineer working in the
United Kingdom for EMI Ltd. He came up with the concept that it was
possible to determine the contents of a box by taking X-ray readings at
multiple positions around the object. The practical computer-based
methodology he subsequently devised to achieve this aim proved to be
consistent with mathematical theory that had been described by the
9781842143094-Ch01 8/8/07 4:26 PM Page 8
Austrian Johann Radon in 1917. At the time of his work, Hounsfield was
unaware of similar developments in the USA by the physicist Allan
McLeod Cormack. In 1979, Hounsfield and Cormack shared the Nobel
Prize for Physiology or Medicine.
Early experiments on a prototype CT scanner built by Hounsfield are
said to have included CT imaging of a cow’s head obtained from a
butcher’s shop.4 Initial images were disappointing in that they visualized
none of the details of brain structure, including the ventricles. His co-
worker, Ambrose, suggested that anatomical detail might have been
obscured by damage to the brain resulting from the blow to the head
used to kill the cow. His theory was proved correct when the experiment
was repeated using a cow’s head from a Kosher butcher, for which the
means of slaughter had been exsanguination. The images obtained on
this occasion displayed the internal brain structure with beautiful clarity.
The first clinical CT system was installed at the Atkinson Morley’s
Hospital in Wimbledon, London. The subsequent expansion of CT into
clinical practice was extraordinarily rapid. The improved visualization
of brain tumors afforded by using contrast media developed for angiog-
raphy and urography was realized very rapidly. Today, along with mag-
netic resonance imaging and ultrasound, the capacity of CT to produce
highly detailed images of the internal structure of the human body is
used not only for diagnostic purposes but also as a valuable adjunct to
the teaching of anatomy in medical education.9
PHYSIOLOGY AND THE CIRCULATION
The science of physiology, the concern for the internal processes and
functioning of the body as opposed to just the structure or anatomy,
really took off as a dynamic new science with William Harvey’s demon-
stration of the circulation of the blood. In fact, several people, namely
Servetus, Colombo, and Cesalpino (16th century Europeans) had
already discovered the pulmonary circulation, but it was Harvey who
became famous for publicizing it to the whole world in his book,
Exercitatio Anatomica de Motu Cordis et Sanguinis in Animalibus (1628),
which for centuries afterward was universally known as ‘de motu
cordis’. This book was written within a few years of another two
famous English books: the King James’ authorized version of the Bible
(1611) and the Folio edition of Shakespeare’s plays (1623). All three
books went down in history as essential reading in their respective
fields4. Harvey’s book was the first significant medical book ever to be
published by an English scholar.
9781842143094-Ch01 8/8/07 4:26 PM Page 9
Harvey was born in 1578 at Folkestone, England, and after his grad-
uate degree at Cambridge he travelled to Padua for his medical train-
ing, where Versalius had held the Chair of Anatomy. He returned to
London at the age of 24 and soon became the royal physician attend-
ing James I. He was described as an aloof man with a displeasing
disposition, but he was deeply respected for his scientific knowledge.
He spent his whole life dissecting as many different animals and people
as he could obtain. Harvey was very careful not to ridicule any concept
of Galen. Through his lectures and vivisections, he painstakingly
worked at repeatedly demonstrating and convincing the English medical
profession of his new ideas before he put them down in writing.
Versalius gave medicine a magnificent anatomical view of the body, but
Harvey built on Versalius’ anatomy to give medicine the full picture of
how the heart worked and how blood moved to bring life to that body.
Although Harvey gave Galen some credit for his thoughts on the circu-
lation, Harvey failed to mention that he himself had been aware of the
advanced ideas of Servetus, Colombo, and Cesalpino.
Quantitative measurements play a fundamental part in physiology,
and, interestingly, the first device for quantifying circulatory parameters
predates Harvey’s work by 25 years. In his work of 1603 entitled Method
vitandorum errorum omnium qui in arte medica contingent [Methods of
avoiding all errors pertaining to the art of medicine], Santorio Santorio
(1561–1636), later Professor of Theoretical Medicine at the University of
Padua, describes an instrument for measuring the pulse rate.10
A jump to 19th century Germany shows us the next important piece
of the jigsaw which helps to make up the complete picture of the devel-
opment of CT perfusion. Adolf Fick (1829–1901) had a remarkable
talent for mathematics and physics, but was persuaded to study medi-
cine by his elder brother, Heinrich. Heinrich, a professor of law, real-
ized that medicine would benefit from Adolf’s talents in other areas.
Soon after completing his medical degree, Fick (Figure 1.2) turned his
attentions to physiology, eventually accepting the Chair of Physiology
at Würzburg. In his Medical Physics,11 Fick introduced profound ideas
on physiological problems such as the mixing of air in the lungs, meas-
uring carbon dioxide output in humans, and the work of the heart. This
book was the first of its kind, and included studies of the hydrodynam-
ics of the circulation. Throughout his life, Fick contributed a steady
stream of information on all three disciplines of mathematics, physics,
and medicine. Even though his major work was on the physiology of
muscle contraction, he used his knowledge to demonstrate how mass
balance could be used to measure cardiac output. The concept, now
known as the Fick Principle, was published in the Proceedings
9781842143094-Ch01 8/8/07 4:26 PM Page 10
Figure 1.2 Adolf Fick, who will be

remembered for expanding physiology to
new dimensions by incorporating mathe-
matics and physics at a new, advanced
level. (Reproduced with permission from
reference 1)
of the Würzburg Physical–Medical Society for July 9, 1870. Interestingly,

the preceding item in the Proceedings announced Röntgen’s election to
the society. As was typical of Fick, he did not attempt to advance or
investigate the proof of his principle, and it took until 1886 for Grehaut
and Quinguad to validate the Fick Principle which forms the basis of
some CT perfusion algorithms in use today.
The localization of tumors on the basis of their circulatory properties
began in the late 1940s. George Moore, a young surgeon in his late 20s
training at Minneapolis, described the use of radioactive diiodofluores-
cein for the diagnosis and localization of brain tumors.12 Interestingly,
Moore’s work was published in the same year as Seymour Kety’s first
description of a non-invasive method to quantify cerebral perfusion in
humans by application of the Fick Principle to measurements of the arte-
rial and venous concentrations of a freely diffusible tracer.13 To this day,
the Kety–Schmidt method remains the basis for tissue perfusion imaging
using [15O]water positron emission tomography and stable xenon CT.
PERFUSION COMPUTED TOMOGRAPHY: COMBINING

ANATOMICAL IMAGING AND QUANTITATIVE PHYSIOLOGY
The anatomical imaging modality of CT and the principles of tracer

theory and physiological imaging began to be combined within a
few years of Hounsfield’s description of CT. The earliest use of CT for
9781842143094-Ch01 8/8/07 4:26 PM Page 11
quantifying vascular physiology comprised attempts to measure cerebral

blood volume from contrast-enhanced CT in the mid-1970s,14,15 but
the physiological models adopted resulted in error due to omitting
the effects of leakage of contrast medium into the extravascular space.
In 1980, Leon Axel, working in the Department of Radiology at the
University of California, described a method that used deconvolution to
determine cerebral blood flow from rapid-sequence contrast-enhanced
CT,16 with the production of parametric images presented the following
year.17 However, at that time the technique was constrained by the slow
speed of image acquisition and data processing of the commercially
available conventional CT systems. Nevertheless, the slower acquisition
protocols required for the estimation of blood–brain barrier (BBB) per-
meability were feasible, and the first CT-derived parametric images of
BBB permeability in human brain tumors were published by Groothius
et al. in 1991.18
The introduction of spiral CT systems in the early 1990s made it
possible to acquire images with the high frequency required for reliable
perfusion imaging. Hitherto, such acquisition protocols had only been
achievable using electron beam CT systems, for which availability was
extremely low. The first report of CT perfusion imaging using a spiral
system was by Miles (Figure 1.3) et al., working at Addenbrooke’s
Figure 1.3 Ken Miles (left), one of the first to implement perfusion computed
tomograpy (CT) on spiral CT systems, receiving from Sir Godfrey Hounsfield
(right), the inventor of CT, a certificate commemorating the 1999 Sir Godfrey
Hounsfield Lectureship of the British Institute of Radiology
9781842143094-Ch01 8/8/07 4:26 PM Page 12
Hospital, Cambridge, UK.19 Calculating perfusion from the maximal rate

of tissue enhancement, the analysis method adopted was based on the
Fick principle rather than the deconvolution approach described by
Axel. In 1992, the Cambridge group presented the first application of
this technique to tumor imaging to the Radiological Society of North
America, demonstrating CT perfusion images of liver tumors, which
were later published in Radiology.20 Within a short time, spiral CT sys-
tems became widely available, to be followed by a similar expansion of
multidetector CT. The ability to use spiral CT for perfusion imaging,
combined with the subsequent release of commercial software to
perform CT perfusion analysis, finally made CT perfusion a practical
technique for the clinical investigation of tumors, as described in the
succeeding chapters of this book.
REFERENCES
1. Fishman AP, Richards DW. Circulation of the Blood: Men and Ideas. New York:
Oxford Univesity Press, 1964.
2. Persaud TVN. Early History of Human Anatomy: From Antiquity to the
Beginning of the Modern Era. Illinois: Charles C Thomas, 1984.
3. Porter R. Clinical science. In: The Greatest Benefit to Mankind: a Medical History
of Humanity from Antiquity to the Present. London: HarperCollins, 1997: 561–96.
4. Friedman M, Friedland GW. Medicine’s 10 Great Discoveries. New Haven: Yale
University Press, 1998.
5. Ferro JM. Egas Moniz (1874–1955). J Neurol 2003; 250: 376–7.
6. Doby T. Victory in an important area: cerebral angiography (1926–1931). In:
Development of Angiography and Cardiovascular Catheterization. Littleton,
MA: Publishing Sciences Group, 1976: 73–93.
7. Hounsfield GN. Computerized transverse axial scanning (tomography).
I. Description of system. Br J Radiol 1973; 46: 1016–22.
8. Ambrose J. Computerized transverse axial scanning (tomography). II. Clinical
application. Br J Radiol 1973; 46: 1023–47.
9. Miles KA. Diagnostic imaging in undergraduate medical education: an expanding
role. Clin Radiol 2005; 60: 742–5.
10. Gedeon A. Science and Technology in Medicine: an Illustrated Account Based on
Ninety-nine Landmark Publications for Five Centuries. New York: Springer, 2006.
11. Fick A. Compendium der Physiologie des Menschen mit Einschluss der
Entwickelungsgeschichte. Wien: Leipzig, 1860.
12. Moore GE. Use of radioactive diiodofluorescein in diagnosis and localization
of brain tumours. Science 1948; 107: 569–71.
13. Kety SS, Schmidt CF. The nitrous oxide method for the quantitative determi-
nation of cerebral blood flow in man: theory, procedure and normal values.
J Clin Invest 1948; 27: 476–83.
14. Penn RK, Walser R, Ackerman L. Cerebral blood volume in man. JAMA 1975;
234: 1154–5.
9781842143094-Ch01 8/8/07 4:26 PM Page 13
15. Zilkha E, Ladurner G, Iliff LD, Du Boulay GH, Marshall J. Computer subtrac-
tion in regional cerebral blood-volume measurements using the EMI-scanner.
Br J Radiol 1976; 49: 330–4.
16. Axel L. Cerebral blood flow determination by rapid-sequence computed
tomography: theoretical analysis. Radiology 1980; 137: 679–86.
17. Berninger WH, Axel L, Norman D, Napel S, Redington RW. Functional imag-
ing of the brain using computed tomography. Radiology 1981; 138: 711–16.
18. Groothius DR, Vriesendorp FJ, Kupfer B et al. Quantitative measurements of
capillary transport in human brain tumours by computed tomography. Ann
Neurol 1991; 30: 581–8.
19. Miles KA, Hayball M, Dixon AK. Colour perfusion imaging: a new application
of computed tomography. Lancet 1991; 337: 643–5.
20. Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion
obtained with dynamic computed tomography. Radiology 1993; 188: 405–11.
9781842143094-Ch01 8/8/07 4:26 PM Page 14
9781842143094-Ch02 8/8/07 11:54 AM Page 15
2
Scientific basis and validation
Ting-Yim Lee and Errol Stewart
INTRODUCTION
In this chapter, the theory of computed tomography (CT) tumor perfu-

sion measurement will be discussed. Because blood attenuates X-rays
uniformly on the scale of the spatial resolution of a CT scanner, flow-
ing blood cannot be differentiated from stationary blood. To measure
tumor perfusion with CT, contrast is injected intravenously, to ‘label’ the
blood. Assuming that the injected contrast is uniformly mixed with
blood, tracing blood through the tumor circulation is equivalent to
tracking a bolus of contrast through the tumor. As such, we can make
use of the extensive literature on tracer kinetics modeling in the meas-
urement of CT tumor perfusion. Note that ‘tumor’ in this discussion can
be broadened to include peritumoral and normal tissue, since the same
consideration would apply in their cases; furthermore we can use the
terms perfusion and blood flow interchangeably. Also, in the diagnosis
of tumor or the study of tumor biology, it is highly advantageous that
besides perfusion we can measure additional functional parameters, as
discussed in the following section, in the same study.
The fundamental processes underlying CT measurement of tumor
perfusion and associated hemodynamic (functional) parameters are the
transport by blood flow of an intravenously administered iodinated
contrast agent to the tumor and exchange by diffusion of these contrast
molecules between the intravascular space and the extravascular inter-
stitial space.1–10 With the current fast CT scanners, both tissue and
vascular contrast concentrations can be measured and traced over time
at short intervals to allow detailed modeling of the distribution of con-
trast agent in tissue. Both compartmental and distributed parameter
models for contrast transport and exchange have been developed to
quantify tumor blood flow, blood volume, mean transit time, and
capillary permeability surface area product.1–10 These parameters as
9781842143094-Ch02 8/8/07 11:54 AM Page 16
well as the models used to measure them will be further discussed in

the following sections.
DEFINITION OF BASIC TERMS AND DISCUSSION

OF FUNDAMENTAL RELATIONSHIPS
In modeling the transport and distribution of X-ray contrast agent in

tissue, the following terms and fundamental relationships need first to
be discussed.
Blood flow (F) This is the volume flow rate of blood through
the vasculature in a tumor. It is usually expressed in units of
ml/min/100g. Note that blood flow measured with CT using the
methodology discussed here includes flow in large vessels, arterioles,
capillaries, venules, and veins. In particular, flow in arteriovenous
shunts will also be included.
Blood volume (Vb) This is the volume of blood within the vascula-
ture in a tumor that is actually ‘flowing’. Again, as in the case of blood
flow, blood volume includes blood in large vessels, arterioles, capillar-
ies, venules, and veins. Any stagnant pool of blood will not be included
in the blood volume. It is measured in units of ml/100g.
Mean transit time (Tm) This is the average time taken by blood ele-
ments to traverse the vasculature from the arterial end to the venous
end in a tumor. If the perfusion pressure (or pressure head) is high,
blood elements are traveling at a higher velocity, resulting in a shorter
mean transit time than when perfusion pressure is low. In this sense,
mean transit time is a surrogate measure of perfusion pressure.11 Mean
transit time is usually measured in seconds.
Distribution volume (Vd) This is defined as the volume within
which the mass of contrast agent in a unit mass tumor would be
distributed at the concentration of the blood with which the tumor is at
equilibrium. If Me is the mass of contrast agent within a component of
the unit mass tumor, for example the interstitial space, and Cb is
the blood concentration, then the distribution volume in the interstitial
space, Ve, is Me/Cb and is in units of ml/100g. Note that the
total distribution volume of contrast agent in the unit mass tumor
Vd = Ve + Vb.
Capillary permeability surface area product (PS) The product of
permeability and blood concentration of a solute gives the unidirec-
tional diffusional flux from blood to interstitial space per unit surface
area of capillary endothelium.11,12 PS is the product of permeability and
the total surface area of capillary endothelium in a unit mass of tumor
9781842143094-Ch02 8/8/07 11:54 AM Page 17
Scientific basis and validation 17
(usually 100g), and hence is the total diffusional flux across all
capillaries. It is measured in units of ml/min/100g. Expressed
in the stated units, PS can be interpreted as the unidirectional flux of
solutes from blood plasma to intersitial space that is equivalent to
the complete transfer of all the solutes in PS milliliters of blood
per minute to the interstitial space. Note that PS is implicitly tied to
the experimental set-up, in which a permeable membrane is separating
two solvents containing a solute at different concentrations. It is
equal to the diffusional flux of solute across the whole permeable
membrane (capillary endothelium) when the solvent (blood) is
‘stationary’ with respect to the membrane. For the more physiological
case of blood flowing through capillaries, the unidirectional
flux of blood-borne solutes through all the capillaries is dependent
on blood flow and PS. This relationship will be discussed in the
following.
Extraction efficiency (fraction) (E) This is the fraction of solutes
present in arterial inlets, with the potential to diffuse into the interstitial
space, that actually becomes transferred from blood to interstitial space
during a single passage of blood from the arterial end to the venous
end of the capillaries of a tumor.13 The mass of solute transferred to
the interstitial space is F ⋅ (Ca −Cv), where Ca is the arterial and Cv is the
venous concentration of the solute. The mass of solute delivered to
the tissue which can diffuse into the interstitial space is F ⋅ (Ca – Ce) until
the arterial concentration approaches the interstitial concentration, Ce.
An operational definition of E is, therefore:
C a − Cv
E= (1)
Ca − C e
Transfer flux of solute through the capillary endothelium

The Johnson and Wilson model is a convenient starting point for this
discussion.14,15 In this model (Figure 2.1), it is assumed that the tumor
consists of capillaries and interstitial tissue. All the capillaries are
lumped together as a single cylinder of length L and volume Vb. The
interstitial tissue is assumed to be a cylindrical annulus around the cap-
illary. As the blood-borne solute enters the capillary, it starts to diffuse
across the capillary endothelium, and thus the blood concentration of
solute Cb will be a function of both axial position, x, along the capil-
lary as well as time, t. The interstitial concentration of solute is Ce(t)
and depends only on time, i.e. the interstitial space is treated as a
9781842143094-Ch02 8/8/07 11:54 AM Page 18
Interstitial space
Ce(t), Ve
Intravascular space
FCa(t) Cb(x,t), Vb FCv(t)
PS
Figure 2.1 The Johnson and Wilson model for the distribution of blood-borne
solutes in tumor. The symbols are explained in the text. Solute concentration in
the intravascular (blood) space, Cb(x,t), is dependent on position along the cap-
illary, to reflect that it is decreasing from the arterial (Ca(t)) to the venous (Cv(t))
end of the capillary. The interstitial space is assumed to be a compartment with
no concentration gradient within it
‘well-stirred’ compartment. The transport and exchange of solute

through the capillaries can be described by the following equation:
∂C b ( x,t ) FL ∂C b ( x,t ) PS
+ + ⎡ C ( x,t ) − Ce ( t ) ⎤⎦ = 0
Vb ⎣ b
(2)
∂t Vb ∂x
For the case when Ce(t) is a constant, say Ce, equation (2) has the
solution:
⎛ V ⎞ − PS x PS
− x ⎛ V ⎞
C b (x,t) = Ca ⎜ t − b x ⎟ e FL + Ce − Ce e FL H ⎜ t − b x ⎟ (3)
⎝ FL ⎠ ⎝ FL ⎠
where H(t) is the unit step function. As expected, at x = 0, Cb(x,t) =

Ca(t), and:
⎛ V ⎞ − PS ⎛ − ⎞
PS
C v ( t ) = C b ( x,t ) x = L,t ≥ Vb = Ca ⎜ t − b ⎟ e F + Ce ⎜ 1 − e F ⎟
F ⎝ F⎠ ⎝ ⎠
Further, the arteriovenous difference can be written as:
⎛ − ⎞
PS
Ca ( t − Tc ) − C v ( t ) = ⎜ 1 − e F ⎟ ⎡⎣ Ca ( t − Tc ) − Ce ⎤⎦
⎝ ⎠
9781842143094-Ch02 8/8/07 11:54 AM Page 19
where Tc = Vb/F is the capillary transit time. Thus:
Ca ( t − Tc ) − C v ( t ) −
PS
E≈ = 1− e F (4)
Ca ( t − Tc ) − Ce
as Crone13 and Renkin16 had previously derived (equation (1)). Moreover,

Fick’s law gives the change in the interstitial concentration as:
dCe ( t )
= F ⎡⎣Ca ( t − Tc ) − C v ( t ) ⎤⎦
dt
which, according to equation (4), can also be expressed as:
dCe ( t )
Ve = FE ⎡⎣Ca ( t − Tc ) − Ce ⎤⎦
dt
The last equation can be interpreted as that the forward flux from the
capillary to the interstitial space is FE·Ca(t−Tc), and the backflux from
the interstial space to the capillary is FE⋅Ce. Thus, FE is the unidirec-
tional flux of solute per unit concentration, or transfer constant, from
blood to interstitial space or from interstitial space to blood. The above
derivation is obtained under the special case when Ce(t) is held con-
stant in time. For the general case when Ce(t) is an arbitrary function of
time, St Lawrence and Lee7 have shown that the unidirectional flux of
solute per unit concentration is still FE.
There exist three regimes for the exchange of solute between
blood and interstitial space: (1) when PS <<F, FE approximates PS,
the exchange is diffusion limited; (2) when PS >>F, so that FE
approaches F, the exchange is flow limited; and, (3) when PS is of
the same magnitude as F, the exchange is neither diffusion nor flow
limited.
Central volume principle

Blood flow, blood volume, and mean transit time are related via the
central volume principle,17 which states that:
Vb
F=
Tm
9781842143094-Ch02 8/8/07 11:54 AM Page 20
This principle applies even for the case of non-intravascular

(diffusible) solutes, provided that the distribution volume is extended
to include the portion of the interstitial space occupied by the solute at
equilibrium.18
X-ray contrast agent In general, this is a derivative of iodobenzoic
acid with a molecular weight of about 700 daltons and is hydrophilic
and inert. Because of the latter properties, it is excluded from the intra-
cellular space in both blood and parenchymal tissue, and only distrib-
utes in blood (plasma) and interstitial (extravascular extracellular)
space. However, for convenience in this chapter, blood flow and
volume are normalized with respect to whole blood. The whole blood
normalized quantities can be converted into plasma normalized quan-
tities by multiplying by the factor: 1 −r·H, where H is the hematocrit
of blood determined using samples from large peripheral vessels,
and r is the ratio of small vessel (at the tissue level) to large vessel
hematocrit.19–22
CT enhancement This refers to the increase in X-ray attenuation in
either blood or tumor in the presence of X-ray contrast agent.
Enhancement is usually measured in Hounsfield units (HU or CT
number), and is related to the concentration of contrast agent present.
The important advantage of CT relative to magnetic resonance (MR)
imaging is that enhancement is linearly related to contrast agent con-
centration and is independent of the microenvironment present at the
tissue level.23 A rule of thumb is that 1 mg of iodine per ml will lead to
an enhancement of around 30 HU when scanning at 80kVp, independ-
ent of how the contrast agent is distributed within the 1-ml volume.23
For CT tumor perfusion studies, both blood and tumor enhancements
are measured in equivalent HU within the same CT image so that cross-
calibration is not necessary.
IMPULSE RESIDUE FUNCTION AND CONVOLUTION
In tracer kinetics modeling, the impulse residue function (IRF)24 and the
mathematical operation of convolution are frequently used to simplify
and provide insight into the dual processes of the convective transport
by blood flow of intravenously administered iodinated contrast agent to
the tumor, and the exchange by diffusion of these contrast molecules
between the intravascular space and the extravascular interstitial space.
In the following, the IRF will be defined and the convolution operation
explained in more detail.
9781842143094-Ch02 8/8/07 11:54 AM Page 21
Impulse residue function

Considering a certain mass of tumor, tumor perfusion (F) carries with it
contrast at a concentration of Ca(t). The tumor time–density curve
(TDC), Q(t), as measured by dynamic CT scanning, is also called the
tumor residue function. In the special case when F·Ca(t) is a delta func-
tion, such that a unit mass of contrast is deposited in the tumor instan-
taneously at time zero as a bolus, then the corresponding tumor residue
function has the general shape as shown in Figure 2.2a. The mathemat-
ical form or numerical representation of the tissue residue function cor-
responding to the instantaneous deposition of unit mass of contrast is
called the impulse residue function (IRF).24 We will give an intuitive
explanation of Figure 2.2 in this section, and a more vigorous (mathe-
matical) justification will be presented later when kinetics modeling is
discussed. There are two distinct phases in the IRF shown in Figure 2.2a.
In the first or vascular phase, there is an abrupt rise, because the con-
trast is injected directly into the arterial input, a plateau of duration that
is equal to the minimum transit time, Tmin, through the vasculature of
the tumor, and a decrease at Tmin. Note that if blood flows as a
‘plug’ through the tumor vasculature, as is assumed in distributed
parameter models14 to simplify the modeling, Tmin is also equal to the
a b
1.2 70
Impulse residue function (no units)
Flow scaled impulse residue

function (ml min−1 (100 g)−1
1 60
50
0.8 Vb
40
0.6
30
0.4 Vb
Vb 20
Tmin = Tm = Tmin = Tm =
F F
0.2 10
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Time (s) Time (s)
Figure 2.2 (a) The impulse residue function according to the Johnson and
Wilson model. The symbols are defined in the text. (b) The blood flow scaled
impulse residue function is obtained by scaling the impulse residue function in
(a) by blood flow, which in this example is 60 ml/min/100 g
9781842143094-Ch02 8/8/07 11:54 AM Page 22
mean transit time (Tm). As the mass of contrast agent injected is unity,
the plateau height in the vascular phase is also unity. The drop at Tmin
represents the fraction of contrast injected that is not extracted by the
tumor and is carried out with the venous blood. The second or intersti-
tial phase begins at Tmin and shows a slow decay towards baseline. The
initial height of the second or interstitial phase is the fraction of
contrast that is extracted by the tumor, or it is equal to the extraction
efficiency of contrast by the tumor. The slowly decaying interstitial
phase represents the return of the extracted contrast from the tumor
to the bloodstream and then clearance via blood flow. The tumor
IRF can be interpreted as the fraction of contrast that remains (‘resides’)
in the tumor after an ‘impulse’ injection of unit mass of contrast, and
as such is unitless. If the amount deposited is M0 instead, then the
corresponding tissue residue function is the product of M0 and the
IRF and in this case M0·IRF has the same units as the tissue
residue function. The IRF is a theoretical concept and cannot be meas-
ured easily in clinical practice since it requires, as a close approxima-
tion, an intra-arterial bolus injection into one of the supply arteries of
the tumor.
Convolution
An intravenous injection of contrast gives rise to an arterial input TDC,
Ca(t), which is not a delta function, and the corresponding tumor
residue function or tumor time–density curve, Q(t), is related to Ca(t)
via the impulse residue function as follows:17
Q(t) = F ⋅ Ca (t) ⊗R(t) = F ⋅ ∫ Ca ( u ) ⋅ R ( t − u ) ⋅ du (5)

0
where ƒ is the convolution operator. The product of F·Ca(u)·∆u gives

the amount of contrast deposited in a small time interval ∆u at time u,
and as explained below with the help of Figure 2.3, the convolution
operation is a generalized multiplication of the mass of contrast
deposited at different times (t′) and the IRF. If blood flow is
unchanged (i.e. stationary) between two identical ‘unity’ bolus injec-
tions as shown in Figure 2.3a, then the tumor TDC corresponding to
each injection is the IRF discussed above, resulting in Figure 2.3b.
In addition, if enhancement measured with a CT scanner is linear
with respect to tumor concentration of contrast, the tumor TDC
9781842143094-Ch02 8/8/07 11:54 AM Page 23
a b
1.2 2.0
Arterial conc. (HU/ml)
Tissue conc. (HU/g)

1.0
1.5
0.8
0.6 1.0
0.4
0.5
0.2
0.0 0.0
0 15 30 45 60 75 90 105 0 15 30 45 60 75 90 105
Time (s) Time (s)
c d
200 10
Tissue conc. (HU/g)

160 8
120 6
80 4
40 2
0 0
0 15 30 45 60 75 90 105 0 15 30 45 60 75 90 105
Time (s) Time (s)
e f
350 30
Tissue conc. (HU/g)
300 25
250 20
200
15
150
10
100
50 5
0 0
0 15 30 45 60 75 90 105 0 15 30 45 60 75 90 105
Time (s) Time (s)
Figure 2.3 Graphical illustrations of the convolution operation involving an

arterial time–density curve (TDC) and impulse residue function (IRF). A blood
flow of 60.0 ml/min/100 g is assumed in the illustrations
corresponding to the case in Figure 2.3c, two bolus injections of dif-

ferent amounts of contrast, is Figure 2.3d. Finally, a general arterial
TDC can be represented as a series of bolus injections, as in Figure 2.3e.
For each of these bolus injections the CT scanner measures a response
that is the product of the IRF, blood flow, and the arterial concentration
9781842143094-Ch02 8/8/07 11:54 AM Page 24
at each of the injection times. The resultant tumor TDC (Figure 2.3f)
in response to the general arterial concentration Ca(t) is the sum of
all the scaled IRFs after they have been shifted in time in accordance
with the times of their corresponding bolus injections. In summary, if
the IRF is known, the corresponding tumor TDC in response to
the arterial TDC, Ca(t), can be obtained as a summation of scaled and
time shifted IRFs. The scale factors and time shifts are given by
F⋅Ca(t) and t. This operation, as illustrated in Figure 2.3f, is called a
convolution.
Flow scaled impulse residue function

and model deconvolution
Equation (5) can be rewritten as:
Q( t ) = Ca ( t ) ⊗ ⎡⎣ F ⋅ R ( t ) ⎤⎦ (5a)
where F·R(t) is the blood flow scaled impulse residue function (FIRF).
Note that whereas IRF is unitless, FIRF has the unit of blood flow:
ml/min/100g. Convolution makes it possible in our discussion to
separate the influence of the details of the administration of contrast
from the effects of the inherent properties of the tumor on the tumor
residue function, Q(t). The arterial input TDC in equation (5a), Ca(t), is
governed by factors such as injection rate and dosage related to the
contrast injection. It is also affected by factors related to the central
hemodynamics of the subject, such as ejection fraction and heart
rate. The blood flow scaled impulse residue function, F⋅ R(t), reflects
tumor blood flow F and other inherent properties of the tumor.
Furthermore, for inert contrast agent, it is possible as shown in the
following section to determine the functional (mathematical) form of
the FIRF in terms of tumor blood flow (F), blood volume (Vb), mean
transit time (Tm), distribution volume (Ve), and capillary permeability
surface area product (PS) or extraction efficiency (E) of contrast agent.
In CT tumor perfusion studies, both Q(t) and Ca(t) are measured, and
the FIRF is convolved with the measured Ca(t) to obtain a predicted
(fitted) tumor residue function, Q̃(t). Model deconvolution of the flow
scaled impulse residue function from the measured Q(t) and Ca(t) deter-
mines values for the set of parameters: F, Vb, Ve, and PS (E) iteratively
by adjusting their values, and hence the corresponding FIRF, and
checking the deviations of Q̃(t) from Q(t) at each iteration until the
minimum is achieved.
9781842143094-Ch02 8/8/07 11:54 AM Page 25
COMPUTED TOMOGRAPHY TUMOR PERFUSION

MEASUREMENT WITHOUT THE NEED FOR
KINETICS MODELING
Before a more thorough discussion of kinetics modeling to derive the

tumor FIRF, a method for CT tumor perfusion measurement that is
independent of the model assumed for the transport and exchange of
contrast between blood and tissue is first discussed in this section. The
method was first proposed by Peters et al. for nuclear medicine applica-
tions,25 and has been adapted by Miles26 for CT perfusion measurements.
Fick principle
Considering a mass of tumor, F is the perfusion, Ca(t) is the contrast
concentration in the arterial inlet(s), and Cv(t) is the contrast concentra-
tion in the venous outlet(s). At time T, the accumulated mass of contrast
from the arterial in-flow, Qin(T), is:
Qin ( T ) = F ⋅ ∫ Ca ( t ) ⋅ dt
0
The accumulated mass of contrast from venous outflow is:
Q out ( T ) = F ⋅ ∫ C v ( t ) ⋅ dt
0
The mass of contrast in the tumor at time T, Q(T), is, by conservation

of mass or the Fick principle:27
Q( T ) = F ⋅ ∫ ⎡⎣ Ca ( t ) − C v ( t ) ⎤⎦ ⋅ dt (6)
0
Equation (6) states that the accumulated mass of contrast in the tumor
over a time period [0, T] is equal to the product of tumor perfusion and
the time integral of the arteriovenous difference in contrast concentration.
No outflow assumption
One immediate simplification is to assume that during the time period
[0, T] the venous concentration, Cv(t), or the venous outflow, is
9781842143094-Ch02 8/8/07 11:54 AM Page 26
equal to zero. This assumption is valid if T is less than the minimum

transit time of the tumor. If the ‘no venous outflow’ assumption holds,
equation (6) simplifies to:26
Q( T ) = F ⋅ ∫ Ca ( t ) ⋅ dt
0
which can be rewritten in a form that facilitates the calculation of tumor

perfusion or blood flow, F, as:
⎡ dQ( t ) ⎤
⎢ dt ⎥ = F ⋅ Ca ( T ) (7)
⎣ ⎦t=T
In particular, the rate of contrast accumulation in tissue dQ(t)/dt will be

maximal when the arterial concentration is at its maximum:
⎡ dQ( t ) ⎤
⎢ dt ⎥ = F ⋅ ⎡⎣ Ca ( t ) ⎤⎦ max (7a)
⎣ ⎦ max
Equation (7a) states that tumor perfusion is the ratio of the maximal rate
of accumulation of contrast in tissue or the maximum slope of Q(t) to
the maximum arterial concentration. For this reason, calculation based
on equation (7a) is also called the maximum slope method. Also, in
order to satisfy the no venous outflow assumption, a relatively high
injection rate (15–20 ml/s) has to be used.28
To avoid the no venous outflow assumption, Cv(t) has to be meas-
ured. This will be possible only if we can identify draining veins. If we
are interested in measuring blood flow of the whole organ, for exam-
ple the brain, then the jugular vein can be used. This technique was
used by Kety and Schmidt to measure global brain blood flow in human
subjects in the 1940s.29 For regional tumor blood flow measurement,
the local draining veins have to be identified and contrast concentration
in them measured. In general, this is not possible. When there is signif-
icant outflow of contrast with venous blood, equation (6) suggests that
the true perfusion would be underestimated by equation (7a).
KINETICS MODELING
Here the discussion is limited to modeling methods that can simultane-

ously measure at least more than one from the set of parameters: blood
9781842143094-Ch02 8/8/07 11:54 AM Page 27
flow, blood volume, mean transit time, capillary permeability surface

area product, and extraction efficiency.
Compartment models
A good example of the use of compartmental models to describe the
distribution of blood-borne solutes in tissue was the study by Blasberg
and Patlak.30,31 Groothuis et al.32,33 were the first to apply compartmen-
tal models to describe the distribution of X-ray contrast agent in tumor,
and used CT to study this distribution in canine and human brain
tumors. Since X-ray contrast agents are hydrophilic, they are excluded
from the intracellular space both in blood and in parenchymal tissue.
Furthermore, because contrast agents are usually inert (i.e. not metab-
olized) in tissue, the modeling of the distribution of contrast agent
in tissue calls for only two compartments, blood (intravascular) and
interstitial space (Figure 2.4). The following equation can be written, by
applying the Fick principle to the interstitial space:
dCe ( t )
Ve = K1 ⋅ C b ( t ) − k 2 ⋅ C e ( t ) (8)
dt
where Cb(t) and Ce(t) are the blood and interstitial concentrations
of contrast agent (solute) respectively. Ve is the distribution volume
of contrast agent in the interstitial space. K1 is the forward transfer
constant from the intravascular space into the interstitial space
and k2 is the backflux constant from interstitial space to intrasvascular
space. As discussed previously, for contrast agent, both the forward and
Capillary
endothelium
K1
Intravascular space Interstitial space

Cb, Vb Ce, Ve
k2
Figure 2.4 Two-compartment model used to describe the distribution of inert

contrast agent in tumor. The symbols are defined in the text
9781842143094-Ch02 8/8/07 11:54 AM Page 28
backward transfer constants are equal to FE. Since the intravascular

space is a compartment, distinction between arterial concentration,
Ca(t), and blood concentration in the space, Cb(t), is not necessary.
The solution of equation (8) leads to an expression for the interstitial
concentration of contrast agent:
∫ C ( u) ⋅ e
− k⋅ ( t−u )
Ce ( t ) = k ⋅ a
du (9)
0
where k = FE/Ve. Since a CT scanner measures tissue enhancement

that has contributions from both the intravascular and the interstitial
space, thus:
∫ C ( u) ⋅ e
− k⋅ ( t−u )
Ce ( t ) = k ⋅ a
du + Ca ( t ) ⋅ Vb (10)
0
where Q(t) is the mass of contrast agent in a unit mass of tissue.

Equation (10) can also be rewritten as:
Q( t ) = Ca ( t ) ⊗ [ FE ⋅ e-kt +V b ⋅ δ( t )] (11)
where δ(t) is the delta function. By comparing equations (5a) and (11),
the FIRF for the two-compartment model shown in Figure 2.4 is:
F ⋅ R ( t ) = FE ⋅ e-kt + V b ⋅δ( t ) (12)
Equation (11) is the operating equation for estimation of the functional

parameters: FE, Ve, and Vb with the use of the two-compartment model
(Figure 2.4) to describe the distribution of X-ray contrast agents in CT
tumor perfusion studies. The estimation can be achieved by a variety of
non-linear regression methods.34 Potential problems in measuring Ca(t)
and the issue of dual input for the liver will be dealt with in a separate
section later.
Some authors have proposed the following simplification to equa-
tion (11) by invoking the assumption that there is no backflux of
contrast agent from the interstitial to the intravascular space,35
which is the case when the size of the distribution volume Ve is very
much larger relative to the transfer constant FE. Under that assumption,
FE is very much smaller compared to Ve, and thus, e−kt =1 for all t.
9781842143094-Ch02 8/8/07 11:54 AM Page 29
Equation (12) for the blood flow scaled impulse residue function then
simplifies to:
F ⋅ R ( t ) = FE ⋅ H( t ) + V b ⋅δ( t ) (12a)
where H(t) is the unit step function and δ(t) is the delta function. The
tumor residue function, Q(t), simplifies to:
t
Q( t ) = FE ∫ Ca ( u )du + Vb⋅ Ca ( t )
0
which can be rearranged as:

t
Q( t ) ∫0 Ca ( u )du (11a)
= FE ⋅ + Vb
Ca ( t ) Ca ( t )
Equation (11a) is the well-known Patlak plot:30,31 if Q(t)/Ca(t) is

t
plotted vs. ∫ C (u)du

0
a
Ca (t), the result is a straight line with a slope of
FE and an intercept of Vb. However, for tumor imaging, it is doubtful
whether the no backflux assumption will be valid in general. Another
drawback of compartmental models is that F and E (PS) cannot be
measured separately because they are determined together as the trans-
fer constant (FE). This is expected because, by assuming the intravas-
cular space to be a well-mixed compartment, all information related to
the convective transport of solute along the capillaries is lost. Although
FE reflects blood flow, it is confounded by the extraction efficiency.
FE can be a relative measure of perfusion within a tumor only if E is
uniform within the tumor. Except when E is known to be close to unity,
FE is not an absolute measure of tumor perfusion.
Distributed parameter models

The impetus for the use of non-compartmental models is to enable the
separation of F and E (PS). Distributed parameter models do not assume
that the intravascular space is a compartment with a spatially uniform
solute concentration. Instead, there is a concentration gradient from the
arterial to the venous end of capillaries to reflect the diffusion of solute
across the capillary endothelium into the interstitial space as blood
travels down the length of capillaries. The simplest distributed parame-
ter model was first proposed by Johnson and Wilson,14 in which the
9781842143094-Ch02 8/8/07 11:54 AM Page 30
interstitial space is still assumed to behave as a compartment (Figure 2.1).

This assumption is justified because capillaries in tissue tend to be ran-
domly oriented so that the arterial and venous ends of different capil-
laries are juxtaposed with respect to each other, leading to a uniform
concentration of solute in the interstitial space when a large ensemble
of capillaries is studied, as would be the case in CT scanning. The gov-
erning equations of the Johnson and Wilson model can be written as:
∂C b ( x,t ) FL ∂C b ( x,t ) PS
+ + ⎡ C ( x,t ) − Ce ( t ) ⎤⎦ = 0
Vb ⎣ b
(13a)
∂t Vb ∂x
dCe ( t ) PS L
L ∫0 ⎣ b
Ve = ⎡C ( x,t ) − Ce ( t ) ⎤⎦ dx (13b)
dt
Equation (13a) describes the convective and diffusional transport of

solute in capillaries and is the same as equation (2), while equation
(13b) gives the rate of change of solute concentration in the interstitial
compartment. However, even with the simplification that the interstitial
space is a compartment, the solution of the Johnson and Wilson model
can only be expressed in the frequency domain with the use of Laplace
transformation.14 This severely limited its application, until St Lawrence
and Lee discovered an adiabatic approximation to derive a closed form
solution of the model in the time domain.7 The motivation for the adia-
batic approximation is twofold. First, the time rate of change of Ce(t) is
much slower than that of Cb(x,t), such that Ce(t) can be approximated
by a staircase function consisting of discrete, finite steps, provided that
the time interval of each step is small relative to the transit time of the
tissue (capillaries). With this approximation, Ce(t) is constant within each
step of the staircase. Second, as discussed above in the solution of equa-
tion (2), when Ce(t) is a constant, Cb(x,t) can be expressed in terms of
Ce(t) by solving equation (13a). Thus, at each step of the staircase
approximation of Ce(t), equation 13a is solved for Cb(x,t) in terms of
Ce(t). With Cb(x,t) expressed in Ce(t), equation (13b) can be used to
determine the increase in Ce(t) at the end of the step. This procedure
can be repeated for each step in the staircase approximation of Ce(t),
resulting in a time domain solution for the mass of solute per unit mass
of tissue, Q(t), which can be expressed in the form of equation (5a) with
blood flow scaled impulse residue function (FIRF), F·R(t), expressed as:7
⎧⎪F 0 < t ≤ Tm
F ⋅ R( t ) = ⎨ (14)
⎪⎩FE ⋅ e ⋅ H( t - Tm ) t > Tm
- k ( t-Tm )
9781842143094-Ch02 8/8/07 11:54 AM Page 31
where Tm = Vb/F according to the central volume principle,

k = FE/Ve, and H(t) is a unit step function. Figure (2b) is a plot of F·R(t)
or the blood flow scaled impulse residue function. It lends itself to the
following interpretation: if a bolus of contrast agent is injected directly
into the arterial inlet of the tumor so that Ca(t) is held at unity concen-
tration for a very short while, the total mass of solute delivered to the
tissue is numerically equal to F. The blood flow scaled impulse residue
function, F·R(t), therefore reaches a height of F immediately and main-
tains this height for a duration equal to the mean transit time (Tm) of
the tissue, Vb/F. The shaded area in Figure (2.2b) is therefore the blood
volume Vb. After a time equal to Vb/F or the mean transit time (Tm),
unextracted contrast agent starts to leave the tissue, F·R(t) drops to a
height of FE, and thereafter contrast agent in the interstitial space back-
diffuses into the intravascular space and is washed out by blood flow.
This portion of F·R(t) is described by a decreasing monoexponential
function with a rate constant equal to FE/Ve. With Ca(t) and Q(t) meas-
ured by CT scanning, the parameters: F, Vb, Ve, and E (PS) in F·R(t) for
the Johnson and Wilson model, equation (14), can be determined by
model deconvolution as discussed above.
INPUT ARTERIAL CONCENTRATION FUNCTION
While the measurement of the mass of contrast agent per unit mass of
tissue, Q(t), is relatively simple because of the inherent linear relation-
ship between CT enhancement and contrast agent concentration,
there are a number of additional factors requiring attention with
respect to the determination of Ca(t). These factors include: (1) partial
volume averaging; (2) possible dispersion of the arterial concentration
function between the site of measurement and the site of input into
the tissue;36 (3) delay in arrival at the input relative to the measure-
ment site;36 and (4) the fact that some tissues can have more than one
input; for example, the liver, besides the hepatic artery input, also
receives blood from the portal vein, which drains venous blood from
the gut.
Partial volume averaging

This arises because the artery selected for the measurement of Ca(t) is
smaller than the resolution limits of CT scanners (typically ~0.7 mm).
Enhancement in a region of interest placed on the artery will be averaged
with that in the surrounding tissue, or partial volume averaging (PVA),
9781842143094-Ch02 8/8/07 11:54 AM Page 32
resulting in the measured arterial concentration, Cam ( t ), being lower

than the true concentration, Ca(t). That is:
Cam ( t ) = k pva ⋅ Ca ( t ) (15)
m
where k is less than or equal to one. When Ca ( t ) instead of Ca(t) is
used in equation (5a), the measured blood flow, Fm, is equal to the true
blood flow divided by kpva (F/kpva). Since kpva is less than one, Fm is
greater than the true F when PVA affects the measurement of Ca(t).
In brain tumor studies, since all of the intracranial arteries that can be
used for Ca(t) are small, PVA is expected to be significant, or kpva is less
than one in most cases. To correct for the effect of PVA is equivalent
to determining a value of kpva. If there exist regions in veins, for
example the superior sagittal sinus, which are free of PVA, then unlike
arterial curves, venous curves, Cv(t), can be measured accurately. Let
h(t) be the transit time spectrum from artery to vein,36 then
1
C v ( t ) = Ca ( t ) ⊗ h( t ) = ⋅ ⎡ Cm ( t ) ⊗ h( t ) ⎤⎦
k pva ⎣ a
The area underneath the vein curve, Av, is:
∞ ∞
1
A v = ∫ C v ( u )du = ⋅ ∫ ⎡⎣C
m
a
( u ) ⊗h( u )⎤⎦ du (16)
0
k pva 0
Since h(t) is a transit time spectrum from artery to vein:
∫ h( u )du = 1
0
and
∞ ∞
∫ ⎡⎣ Ca ( u ) ⊗ h( u ) ⎤⎦ ⋅ du = ∫ Ca ( u ) ⋅ du = A a
m m m
0 0 (17)
m
where A a is the area underneath the measured arterial curve. From
equations (16) and (17), k is determined as the ratio:
A am
k pva =
Av
9781842143094-Ch02 8/8/07 11:54 AM Page 33
Dispersion of measured arterial concentration

function relative to true input
To minimize the effect of partial volume averaging, the artery chosen
for the measurement of arterial concentration function is upstream to,
and hence larger in size than, the input to the tissue. Dispersion refers
to the spreading of the arterial input curve at the input to a tissue region
relative to the measurement site, which is upstream to the tissue region.36
The dispersion can be characterized by the transit time spectrum, h(t),
between the measurement site and the tissue region.36
Let Cam ( t ) be the measured arterial curve, and then the arterial curve
at the input to the tissue region can be expressed as:
Ca ( t ) = Cam ( t ) ⊗ h( t ) (18)
Substituting equation (18) into equation (5a), the operating equation for
model deconvolution:
Q(t) = Cam (t) ⊗ h( t ) ⊗ ⎡⎣ F ⋅ R ( t ) ⎤⎦ (19)
Equation (19) suggests that model deconvolution between the measured

arterial and tissue curves gives h(t) b [F◊R(t)] instead of just the blood flow
scaled impulse residue function, F◊R(t). If h(t) can be approximated by
a delta function, the effect of dispersion on the estimation of the param-
eters: F, Vb, Ve and PS (E) in F◊R(t) is minimized since:
h( t ) ⊗ ⎡⎣F ⋅ R ( t )⎤⎦ = F ⋅ R ( t ) when h( t ) ≈ δ ( t )
St Lawrence and Lee,8 using phantom experiments that reproduced the

flow conditions of the human vascular system, demonstrated that the
dispersion effect was negligible (i.e. h(t) is well represented by a delta
function), even for a distance of 90 cm between the measurement and
the input site. Further investigations in animals or patients are required
to resolve this important issue.
Delay in arrival of contrast agent in tissue relative to the

measured arterial concentration function
Since the measurement site is upstream to the input, there will be a delay
in the arrival of contrast agent at the tissue relative to the measured
9781842143094-Ch02 8/8/07 11:54 AM Page 34
arterial concentration function. To account for the finite transit time, T0,
between the measurement region and the input to the tumor:
Ca ( t ) = Cam ( t − T0 )
Substituting into equation (5a):
Q( t ) = Cam ( t − T0 ) ⊗ ⎡⎣ F ⋅ R ( t ) ⎤⎦ = Cam ( t ) ⊗ ⎡⎣ F ⋅ R ( t − T0 ) ⎤⎦ (20)
by manipulation of the convolution integral. Equation (20) suggests the

obvious fact that the finite transit time (T0) between the measurement
site for the arterial concentration and the input to the tumor is equiva-
lent to a delay of the same magnitude in arrival in the tumor relative to
the measured arterial TDC, Cam ( t ) . From this latter viewpoint, there is
an additional arrival time (T0) parameter in the blood flow scaled
impulse residue function for each of the models we have discussed
above. Accordingly:
• for the two-compartment model:
F ⋅ R ( t − T0 ) = FE ⋅ e ( 0 ) + Vb ⋅ δ ( t − T0 )
− k t−T
(21)
• for the two-compartment model under the assumption that there is

no backflux of contrast agent from the interstitial to the intravascular
space:
F ⋅ R ( t − T0 ) = FE ⋅ H( t − T0 ) + V b ⋅ δ( t − T0 ) (22)
• for the Johnson and Wilson model:
⎧0 0 ≤ t ≤ T0
⎪
F ⋅ R ( t − T0 ) = ⎨F T0 ≤ t ≤ T0 + Tm (23)
⎪ -k( t-T0-Tm )
⎩FE ⋅ e ⋅ H( t − T0 − Tm ) t > T0 + Tm
Hepatic artery and portal vein input to the liver

The liver receives its blood from both the hepatic artery delivering oxy-
genated blood from the heart and the portal vein draining venous
blood from the gastrointestinal tract.37 Approximately two-thirds of the
9781842143094-Ch02 8/8/07 11:54 AM Page 35
blood flow to the liver is supplied by the portal vein, and the remain-
ing one-third is supplied by the common and proper hepatic arteries.38
The hepatic artery input, Ch(t), can be approximated by the aortic input
at the level of the liver, while the portal vein input, Cp(t), can be directly
measured if one of the CT slices includes the portal vein in its field of
view. If α is the fraction of total blood flow to liver tissue that arises
from the hepatic artery (or hepatic arterial fraction), then Ca(t) can be
expressed as the weighted sum of Ch(t) and Cp(t):39
Ca ( t ) = α ⋅ Ch ( t ) + (1 − α ) ⋅ Cp ( t ) (24)
By replacing Ca(t) in equations (5a), (21), and (23) with that expressed
in equation (24), the hepatic arterial fraction can be estimated together
with other parameters in the two-compartment and Johnson and Wilson
models.
Recirculation
The effect of recirculation on the kinetics modeling proposed above
can be analyzed in a straightforward manner as follows. To account
for recirculation, the arterial TDC, Ca(t), can be written as a summation
of the first pass component, C FP a (t), and subsequent recirculation
components. For simplicity we will assume that only the first recircula-
tion component, C aR1(t), is important, and the following derivation
can easily be generalized to more than one recirculation component.
If tumor perfusion is unchanged (stationary) for the duration of the
first pass and subsequent recirculation phases, then equation (5a)
would apply to all phases with the same blood flow scaled impulse
residue function, F·R(t). The tissue residue function for the first
pass phase, QFP(t), and the first recirculation phase, QR1(t), can be
written as:
QFP (t) = CaFP (t) ⊗⎡⎣F ⋅ R(t) ⎤⎦ (25a)
QR1(t) = CaR1 (t) ⊗⎡⎣F ⋅ R(t) ⎤⎦ (25b)
Adding equations (25a) and (25b) together, we have:
QFP (t) + QR1 ( t ) = ⎡⎣Ca FP (t) + CaR1 ( t )⎤⎦ ⊗ ⎡⎣F ⋅ R(t) ⎤⎦ (26)
9781842143094-Ch02 8/8/07 11:54 AM Page 36
Since, by definition:
Q(t) = Q FP (t) + Q R1 ( t )
Ca (t) = CaFP (t) + CaR1 ( t )
equation (26) can be simplified into:

Q(t) = Ca (t) ⊗ ⎡⎣ F ⋅ R(t) ⎤⎦
which is the same as equation (5a). Thus, provided that blood flow is
stationary, the blood flow scaled IRF, F·R(t), can be calculated by model
deconvolution between the measured arterial and tissue TDCs, which
contain both the first pass and subsequent recirculation phases. There
is no need to correct for recirculation in the measured TDCs.
PRACTICAL ISSUES
In this section, a number of issues concerning the practical implemen-

tation of CT tumor perfusion are discussed.
Scanning protocol: image interval, scanning

duration, and radiation dose
The image interval and scanning duration used in CT tumor perfusion
studies have to be optimized with respect to the model, either two-
compartment or the Johnson and Wilson model, employed in the
description of the blood flow scaled impulse residue function (FIRF),
that is, equation (21), (22), or (23).
For the two-compartment model, equation (21) shows that the FIRF is
a slowly decaying exponential with a half-life measured in minutes.
Hence, the image interval required can be as long as 3–15s and the
duration of scanning required a few minutes, provided that it is at least
3–5 times the half-life of the FIRF. Since current CT scanners can acquire
an image in less than a second, this protocol requires intermittent scan-
ning, which reduces the radiation dose to the patient, or the protocol can
be modified to increase the anatomical coverage of a study.40 The disad-
vantage is, as discussed above, that F and E (PS) cannot be measured sep-
arately because they are determined together as the transfer constant (FE).
9781842143094-Ch02 8/8/07 11:54 AM Page 37
For the Johnson and Wilson model, the blood flow scaled impulse
residue function has two distinct phases. The first or vascular phase,
as shown in Figure 2.2, is relatively brief, being equal to the mean tran-
sit time through the tumor vasculature, which is in seconds rather than
minutes. The second or interstitial phase is a slow exponential decay
with a half-life which is usually measured in minutes. A two-phase
scanning protocol is required for the Johnson and Wilson model. The
first phase of scanning has an image interval of 0.5–1 s and a duration
of 30–45 s to allow reliable determination of the vascular phase of
the FIRF. The short image interval in the first phase requires the X-ray
to be on all the time, to scan continuously, and as such constitutes
the bulk of the radiation dose to the patient. The brevity of the
duration of the first phase means that breath-holding required for
studies affected by breathing motion during this phase of scanning is
possible. The second phase of scanning has an image interval of
10–20 s and duration of several minutes to characterize the rate
constant, k, in equation (23). At this infrequent image interval, X-rays are
turned on intermittently to acquire a scan. Stewart et al. have proposed
an intermittent scanning method for this second phase that allows the
correction of breathing motion without respiratory gating. For reasons
of space, interested readers are referred to their original publication.39
The following is a typical two-phase scanning protocol, as an
example:
• first phase: image interval 1 s and 30 s duration
• a delay of 15 s between first and second phases
• second phase: image interval 15 s and 2 min duration.
At each CT slice location, the first phase acquires 30 images
while the second phase acquires only eight images. Thus, the first
phase contributes close to 80% of the total radiation dose if the
same X-ray tube current (mA) is used in both phases. We can also
compare the radiation dose required by the two-compartment and
Johnson and Wilson models. For the same duration of 2.5 min and at
15s image interval, the two-compartment model protocol acquires
10 images per slice location while the Johnson and Wilson protocol
acquires 38. The radiation dose for the Johnson and Wilson
model scanning protocol is 3.8 times higher than that of the two-
compartment model, again assuming that the same X-ray tube current
is used to acquire each image. On the other hand, the advantage of the
Johnson and Wilson model is that absolute tumor perfusion can be
measured.
9781842143094-Ch02 8/8/07 11:54 AM Page 38
Calibration of CT scanner
In CT tumor perfusion studies, there is an implicit assumption that
enhancement measured by a CT scanner in HU in artery and tumor is
linearly proportional to their contrast concentration, Ca(t) and Q(t) in
equation (5a). As long as the proportional relationship is the same for
both artery and tumor, calibration of enhancement in Hounsfield units
(HU) versus contrast concentration is not required. This is true for
CT scanners when beam hardening is negligible. For brain tumor per-
fusion studies, given the relatively low enhancement in arteries and
veins in the brain at the dosage of contrast injected,41 this condition is
usually satisfied. For other tissues, particularly in the abdominal or tho-
racic region, the dosage of contrast injected may have to be limited to
reduce the beam hardening effect arising from the superior vena cava
or the aorta, if the injection site is any one of the upper-extremity veins.
Contrast, noise, and radiation dose

The quality of the arterial and tissue TDCs, used to derive perfusion and
related parameters with either the maximum slope method or model
deconvolution method, depends on their signal-to-noise ratio. The
signal is the arterial and tissue contrast enhancement detected by a
CT scanner. Since all CT contrast agents are iodinated compounds, the
main process contributing to contrast enhancement is the absorption of
X-ray photons at the energy of the K-edge of iodine, or 33.2keV. A low
kVp X-ray beam would increase the signal (enhancement) but at the
expense of noise. As demonstrated by Lee et al. with a brain-size phan-
tom complete with a bone ring to simulate the skull, increasing kVp
from 80 to 120, but keeping the mAs used to acquire a CT image the
same, increases the signal-to-noise ratio by 11%, but increases the radi-
ation dose to the subject by over 300%.42 Thus, when signal-to-noise
performance is normalized with respect to radiation dose imparted,
80kVp is better than 120kVp for CT perfusion studies. Essentially the
same conclusion was reached in studies by Wintermark et al.43 and
Hirata et al.,44 in which patients were scanned at both 80 and 120kVp,
and signal-to-noise ratios as well as radiation doses were compared.
The same conclusion was arrived at by Huda et al. using a body-sized
phantom: when mAs was kept the same but kVp increased from 80 to
120kVp, the signal-to-noise ratio increased by less than 1.2-fold while
the radiation dose increased by over 3-fold.45 Thus, 80 kVp is the
optimal kVp to be used in CT tumor perfusion studies, regardless of
whether the tumor is located in the brain or abdominal organs.
9781842143094-Ch02 8/8/07 11:54 AM Page 39
VALIDATION OF COMPUTED TOMOGRAPHY TUMOR

BLOOD FLOW AND RELATED MEASUREMENTS
IN ANIMAL TUMOR MODELS
The accuracy and precision of CT tumor blood flow and related meas-
urements has been investigated in animal models of brain tumor, soft
tissue tumor, and liver tumor, using methods that do not require mod-
eling, as well as kinetics modeling methods. These preclinical studies
have the advantage that the accuracy of CT measurements can be
validated against reference gold-standard microsphere measurements,
which require tissue sampling and hence sacrifice of the animal. In this
section, the results from these validation studies are discussed.
Methods not requiring tracer kinetics modeling

Pollard et al.46 compared CT tumor perfusion measurements obtained
by application of the Fick principle with the no outflow assumption
(equation (7a)) against those measured with the gold-standard (fluores-
cent) microsphere method47 in male Fischer rats implanted with
R3230AC rat-derived mammary adenocarcinoma tumor in both caudal
thigh regions. The regression equation between the two types of
perfusion measurements was:
Microsphere perfusion = 2.06 × CT perfusion– 0.221

(27)
(r2 = 0.68, p << 0.001)
Both microsphere and CT perfusion values are in ml/min/g. Equation

(27) shows that the true tumor perfusion as measured by microspheres
was underestimated by the Fick principle and no outflow assumption.
This suggests that the assumption of no venous outflow is not valid in
this rat tumor model.
Tracer kinetics modeling methods

R3230AC rat-derived mammary
adenocarcinoma tumor in rat thigh
Pollard et al. also used a two-compartment model (Figure 2.4) under
the assumption that there is no backflux of contrast agent from the
interstitial to the intravascular space, that is, equation (11a), to estimate
the FE (PS) of R3230AC rat-derived mammary adenocarcinoma tumor
in rat thigh.46 As discussed in the section on ‘Transfer flux of solute
9781842143094-Ch02 8/8/07 11:54 AM Page 40
through the capillary endothelium’ above, the FE estimated from

equation (11a) can be equated to the PS product of the tumor only
when PS is much smaller than blood flow (F). In Pollard et al.’s study,
blood flow was comparable to the estimated FE, and thus the latter
cannot be regarded as equal to PS. Nevertheless, their FE values
correlated with the amount of Evans blue dye in the tumor following
2 minutes of circulation of the dye,46 which according to equation (11a)
is a semiquantitative measure of PS.48
VX2 brain tumor

Nine New Zealand White rabbits with implanted VX2 tumors in the
brain were used in this validation study. Measurements of tumor blood
flow, blood volume, and permeability surface area product in the
tumor, peritumor, and contralateral normal tissue regions were
obtained with the Johnson and Wilson model (equations (5a) and
(23)).10 In all nine rabbits, CT-derived blood flow values were com-
pared with those simultaneously obtained by the ex vivo microsphere
technique. In three of the nine rabbits, the variability of repeated blood
flow and blood volume measurements was examined.
There was a significant linear correlation (r = 0.847) between the
CT- and microsphere-derived blood flow values, with a slope not sig-
nificantly different from unity (0.99 ± 0.03, p < 0.01). The mean differ-
ence between blood flow measurements obtained using both methods
did not significantly deviate from zero (p >0.10). The variability in
CT blood flow and blood volume measurements in the repeated
studies was 13% and 7%, respectively.
VX2 skeletal muscle (thigh) tumor

VX2 tumor cells were implanted in the left thigh of nine New Zealand
White rabbits and left to grow to 0.4 cm in diameter. Blood flow (BF),
blood volume (BV), mean transit time (MTT) and PS in the implanted
tumor were measured using the Johnson and Wilson model.49 The
ex vivo method of radioactive microspheres was used to validate
the CT blood flow measurements. The precision of the CT technique
for the measurement of BF, PS, BV, and MTT was determined by
repeating the scan three times in four of the nine rabbits under steady-
state conditions. There was a significant linear correlation (r = 0.96)
between CT- and microsphere-measured blood flow values, with a
slope not significantly different from unity (0.98 ±0.02, p < 0.0001). The
precision of blood flow, capillary permeability surface area product,
9781842143094-Ch02 8/8/07 11:54 AM Page 41
blood volume, and mean transit time was 14%, 18%, 20% and 24%,
respectively.
VX2 liver tumor

VX2 carcinoma cells were implanted into the livers of eight male New
Zealand white rabbits. Each CT tumor perfusion study started with a 30-s
cine scan, with breath-hold at the injection of 5ml contrast, followed by
4-s cine scans without breath-hold every 10s for 2min. These latter images
were registered with the breath-hold images to eliminate breathing
motion.39 Total liver blood flow and hepatic arterial fraction were calcu-
lated according to equations (5a), (23), and (24). Hepatic artery blood
flow (HABF) was further calculated as the product of total liver blood flow
and hepatic arterial fraction. The HABF measured by CT in normal tissue
and tumor rim and core were compared with those measured by the
ex vivo microsphere technique under normocapnia, hypercapnia,
and hypocapnia conditions. The ex vivo microsphere HABF measure-
ments under normal conditions were: 47 ± 24, 35 ± 12, and
95±21 ml min/100g in normal tissue, tumor core, and rim, respectively. In
comparison, CT perfusion HABF measurements were: 54±13, 35±14, and
122±25 ml min/100g in the same regions, respectively. No significant
differences were found between HABF measurements obtained with the
two techniques (p>0.05). Furthermore, there was a strong correlation
when HABF values from the two techniques were compared:
slope of 1.12, intercept of 6.94, and r2 =0.76. The Bland–Altman plot50 com-
paring CT perfusion and microsphere measurements gave a mean differ-
ence of −13.34mlmin−1 (100g)−1. The limits of agreement, the interval in
which 95% of the differences lay, were –28.43 and 55.11 ml min/100g.
The reproducibility of CT perfusion measurements in liver tumor
was investigated in another five male New Zealand White rabbits
(3.1–3.4kg) implanted with VX2 liver tumors. Each rabbit was scanned
twice, 15min apart, with the same protocol as in the validation studies
discussed above. Total hepatic blood flow (HBF), hepatic blood
volume (HBV), hepatic arterial fraction (HAF), contrast arrival time (T0),
and permeability surface area product (PS) were determined as
described above. Data from the tumor core, tumor rim, and adjacent
normal tissue were analyzed using a repeated measures analysis of vari-
ance (ANOVA).51 The reproducibility was assessed by the coefficient of
variation (CV). In all cases, the ANOVA showed no significant differ-
ence between perfusion parameter measurements made in the two
consecutive CT perfusion scans. Table 2.1 shows the CV of the five
parameters assessed.
9781842143094-Ch02 8/8/07 11:54 AM Page 42
Table 2.1 Reproducibility of computed tomography measurements
Coefficient of variation (%)
T0 HBF HBV PS HAF
Tumor rim 17.6 12.7 16.5 20.7 12.9

Tumor core 17.9 10.2 9.6 16.4 7.2
Normal tissue 8.2 8.8 7.6 65.4 16.0
T0, contrast arrival time; HBF, hepatic blood flow; HBV, hepatic blood volume; PS, permeability surface
area product; HAF, hepatic arterial fraction
VALIDATION OF CT TUMOR BLOOD FLOW AND RELATED

MEASUREMENTS IN CLINICAL STUDIES
Reproducibility studies
Goh et al.52 investigated the reproducibility of measurements of blood
flow, blood volume, mean transit time, and permeability surface area
product obtained with the Johnson–Wilson model (equations (5a)
and (23)) in 10 patients with colorectal cancer. Each patient underwent
two single-phase 65-s perfusion studies separated by 48 hours to allow
assessment of reproducibility of the CT measurements listed above. The
mean difference (95% limits of agreement) for blood volume, blood flow,
mean transit time, and permeability when a 5-mm slice thickness was used
were 0.04 (−2.50 to +2.43)ml/100g +8.80 (−50.5 to +68.0)ml/min/100 g;
–0.99 (−8.19 to +6.20)s; and +1.20 (−5.42 to +7.83)ml/min/100 g,
respectively. Similar reproducibility results were obtained for a 20-mm
slice thickness.
Ng et al.53 investigated the reproducibility of measurements of blood
volume and permeability surface area product obtained using a two-
compartment model (Figure 2.4) under the assumption that there is no
backflux of contrast agent from the interstitial to the intravascular space,
that is, equations (5a) and (22). As discussed before, the FE estimated
from equations (5a) and (22) can be equated to the PS of the tumor
only when PS is much smaller than blood flow (F). In their study, blood
flow was not estimated, and thus it is not certain whether the estimated
FE is equal to PS. Nevertheless, the mean difference (95% limits of
agreement) for tumor FE was 1.4 (−4.0 to 6.8)ml/min/100 ml for 10-mm
slice thickness and 0.8 (−3.6 to 5.2)ml/min/100 ml for 40-mm slice
thickness. The mean difference (95% limits of agreement) for blood
volume was 1.9 (−5.1 to 8.9)ml/100ml for 10-mm slice thickness and
9781842143094-Ch02 8/8/07 11:54 AM Page 43
1.4 (−3.7 to 6.6)ml/100 ml for 40-mm slice thickness. The coefficient

of variation for permeability was 18.7% for 10-mm slice thickness,
improving to 11.9% for 40-mm slice thickness. The coefficient of
variation for blood volume was 41.7% for 10-mm slice thickness,
improving to 32.6% for 40-mm slice thickness.
Accuracy studies
Hattori et al.54 compared CT tumor perfusion measurement obtained by
application of the Fick principle with the no outflow assumption
(equation (7a)) against that measured with positron emission tomogra-
phy using the [15O-]carbon dioxide (C15O2) steady-state method in 16
patients with superficial tumors including the sites of larynx, hypo-
pharynx, tongue, lung, and breast. The CT blood flow measurement on
average was about 20% higher than the PET measurement, and the cor-
relation coefficient between the two sets of measurements was high
(r=0.79). These results suggest that CT tumor blood flow measurement
in patients using the no venous outflow assumption is accurate, albeit
with a systemic overestimation of about 20%.
CONCLUSION
A CT tumor perfusion study can be performed using very simple pro-

cedures, making it particularly suited for incorporation into the routine
diagnostic imaging protocol of cancer patients. The kinetics modeling
discussed in this chapter permits simultaneous quantitative determina-
tion of a number of functional parameters reflecting the activity of
angiogenesis in tumors: blood flow, blood volume, mean transit time,
capillary permeability surface area product, and arrival time of contrast
agent, from a single study. For the special case of the liver, an addi-
tional parameter determined is hepatic arterial fraction of the liver
blood flow. Furthermore, measurements of these parameters have been
shown to be accurate and reproducible. With the high spatial resolu-
tion of CT scanning, each of these functional parameters can be deter-
mined and displayed as functional maps in voxels as small as 10mm3
(1.5mm×1.5 mm×5mm). This would allow visualization of the inher-
ently very heterogeneous angiogenic activities seen in tumors. The
utility of imaging angiogenesis has been demonstrated in a number of
experimental studies, and further clinical studies will prove whether the
favorable results obtained in animal models can be reproduced in
human subjects.
9781842143094-Ch02 8/8/07 11:54 AM Page 44
REFERENCES
1. Axel L. Cerebral blood flow determination by rapid-sequence computed

tomography; theoretical analysis. Radiology 1980; 137: 679–86.
2. Axel L. Tissue mean transit time from dynamic computed tomography by a
simple deconvolution technique. Invest Radiol 1983; 18: 94–9.
3. Miles KA. Measurement of tissue perfusion by dynamic computed tomography.
Br J Radiol 1991; 64: 409–12.
4. Blomley MJK, Coulden R, Bufkin CRT et al. Contrast bolus dynamic computed
tomography for the measurement of solid organ perfusion. Invest Radiol 1993;
28 (Suppl 5): S72–7.
5. Blomley MJK, Dawson P. Bolus dynamics: theoretical and experimental
aspects. Br J Radiol 1997; 70: 351–9.
6. Dawson P, Blomley MJK. Contrast agents as extra-cellular fluid space markers:
adaptation of the central volume theorem. Br J Radiol 1996; 69: 717–22.
7. St Lawrence K, Lee T-Y. An adiabatic approximation to the tissue homogeneity
model for water exchange in the brain I. Theoretical derivation. J Cereb Blood
Flow Metab 1998; 18: 1365–77.
8. St Lawrence K, Lee T-Y. An adiabatic approximation to the tissue homogeneity
model for water exchange in the brain II. Experimental validation. J Cereb
Blood Flow Metab 1998; 18: 1378–85.
9. Cenic A, Nabavi DG, Craen RA, Gelb AW, Lee T-Y. Dynamic CT measurement of
cerebral blood flow: a validation study. AJNR Am J Neuroradiol 1999; 20: 63–73.
10. Cenic A, Nabavi DG, Craen RA, Gelb AW, Lee T-Y. A CT method to measure
hemodynamics in brain tumors: validation and application to cerebral blood
flow maps. AJNR Am J Neuroradiol 2000; 21: 462–70.
11. Schumann P, Touzani O, Young AR et al. Evaluation of the ratio of cerebral
blood flow to cerebral blood volume as an index of local cerebral perfusion
pressure. Brain 1998; 121: 1369–79.
12. Rapoport SI. Blood-Brain Barrier in Physiology and Medicine. New York:
Raven Press, 1976.
13. Crone C. The permeability of capillaries in various organs as determined by
use of the ‘indicator diffusion’ method. Acta Physiol Scand 1963; 58: 292–305.
14. Johnson JA, Wilson TA. A model for capillary exchange. Am J Physiol 1966;
210: 1299–303.
15. Sawada Y, Patlak CS, Blasberg RG. Kinetic analysis of cerebrovascular trans-
port based on indicator diffusion technique. Am J Physiol 1989; 256: H794–812.
16. Renkin EM. Transport of potassium-42 from blood to tissue in isolated mam-
malian skeletal muscle. Am J Physiol 1959; 197: 1205–10
17. Meier P, Zierler KL. On the theory of the indicator dilution method for meas-
urement of blood flow and volume. J Appl Physiol 1954; 6: 731–44.
18. Robert GW, Larson KB, Spaeth EE. The interpretation of mean transit time
measurements for multiphase tissue systems. J Theor Biol 1973; 39: 447–75.
19. Brooks DJ, Thomas DG, Marshall J, Jones T. Studies on regional cerebral pH
in patients with cerebral tumours using continuous inhalation of 11CO2 and
positron emission tomography. J Cereb Blood Flow Metab 1986; 6: 529–35.
20. Brudin LH, Valind SO, Rhodes CG, Turton DR, Hughes JM. Regional lung
hematocrit in humans using positron emission tomography. J Appl Physiol 1986;
60:1155–63.
9781842143094-Ch02 8/8/07 11:54 AM Page 45
21. Okazawa H, Yonekura Y, Fujibayashi Y et al. Measurement of regional cere-

bral plasma pool and hematocrit with copper-62-labeled HSA-DTS. J Nucl Med
1996; 37: 1080–5.
22. Sakai F, Igarashi H, Suzuki S, Tazaki Y. Cerebral blood flow and cerebral
hematocrit in patients with cerebral ischemia measured by single-photon
emission computed tomography. Acta Neurol Scand 198; 127 (Suppl): 9–13.
23. Lee TY, Ellis RJ, Dunscombe PB et al. Quantitative computed tomography
of the brain with xenon enhancement: a phantom study with the GE9800
scanner. Phys Med Biol 1990; 35: 925–35.
24. Bassingthwaighte JB, Chinard FP, Crone C, Lassen NA, Perl W. Definitions and
terminology for indicator dilution methods. In: Crone C, Lassen NA, eds. Capillary
Permeability. Copenhagen: Munskgaard 1970: 665–9.
25. Peters AM, Gunasekera RD, Henderson BL et. al. Noninvasive measurement of
blood flow and extraction fraction. Nucl Med Commun 1987; 8: 823–37.
Br J Radiol 1991; 64: 409–12.
27. Miles KA, Charnsangavej C, Lee FT et al. Application of CT in the investigation
of angiogenesis in oncology. Acad Radiol 2000; 7: 840–50.
28. Klotz E, Konig M. Perfusion measurements of the brain: using dynamic CT for
the quantitative assessment of cerebral ischemia in acute stroke. Eur J Radiol
1999; 30: 170–84.
29. Kety SS, Schmidt CT. The determination of cerebral blood flow in man by
the use of nitrous oxide in low concentrations. Am J Physiol 1945; 143:
53–61.
30. Patlak CS, Blasberg RG, Fenstermacher JD. Graphical evaluation of blood-to-brain
transfer constants from multiple-time uptake data. J Cereb Blood Metab 1983;
3: 1–7.
31. Patlak CS, Blasberg RG. Graphical evaluation of blood-to-brain transfer constants
from multiple-time uptake data. Generalizations. J Cereb Blood Metab 1985;
5: 584–90.
32. Groothuis DR, Lapin GD, Vriesendorp FJ, Mikhael MA, Patlak CS. A method to
quantitatively measure transcapillary transport of iodinated compounds in
canine brain tumors with computed tomography. J Cereb Blood Flow Metab
1991; 11: 939–48.
33. Groothuis DR, Vriesendorp FJ, Kupfer B et al. Quantitative measurements of
capillary transport in human brain tumors by computed tomography. Ann
Neurol 1991; 30: 581–8.
34. Gill PE, Murray W, Wright MH. Practical Optimization. London: Academic
Press, 1981.
35. Miles KA. Tumour angiogenesis and its relation to contrast enhancement on
computerised tomography: a review: Eur J Radiol 1999; 30: 198–205.
36. Calamante F, Gadian DG, Connelly A. Delay and dispersion effects in dynamic
susceptibility contrast MRI: simulations using singular value decomposition.
Magn Reson Med 2000; 44: 466–73.
37. Richardson PD, Withrington PG. Liver blood flow. I. Intrinsic and nervous
control of liver blood flow. Gastroenterology 1981; 81: 159–73.
38. Schenk W Jr, McDonald J, McDonald K, Drapanas T. Direct measurement of
hepatic blood flow in surgical patients: with related observations on hepatic
flow dynamics in experimental animals. Ann Surg 1962; 156: 463–71.
9781842143094-Ch02 8/8/07 11:54 AM Page 46
39. Stewart EE, Chen X, Hadway J, Lee T-Y. Correlation between hepatic tumor
blood flow and glucose utilization in a rabbit liver tumor model. Radiology
2006; 239: 740–50.
40. Roberts H, Roberts T, Lee T-Y, Dillon W. Dynamic, contrast-enhanced CT of
human brain tumors: quantitative assessment of blood volume, blood flow,
and microvascular permeability. Report of Two Cases. AJNR Am J Neuroradiol
2002; 23: 828–32.
41. Nabavi DG, LeBlanc LM, Baxter B et al. Monitoring cerebral perfusion flow
after subarachnoid hemorrhage using CT. Neuroradiology 2001; 43: 7–16.
42. Lee TY, Ellis RJ, Dunscombe PB et al. Quantitative computed tomography of
the brain with xenon enhancement: a phantom study with the GE9800
scanner. Phys Med Biol 1990; 35: 925–35.
43. Wintermark M, Maeder P, Verdun FR et al. Using 80 kVp versus 120 kVp in
perfusion CT measurement of regional cerebral blood flow. AJNR Am
J Neuroradiol 2000; 21: 1881–4.
44. Hirata M, Sugawara Y, Fukutomi Y et al. Measurement of radiation dose in
cerebral CT perfusion study. Radiat Med 2005; 23: 97–103.
45. Huda W, Scalzetti EM, Levin G. Technique factors and image quality as functions
of patient weight at abdominal CT. Radiology 2000; 217: 430–5.
46. Pollard RE, Garcia TC, Stieger SM et al. Quantitative evaluation of perfusion
and permeability of peripheral tumours using contrast-enhanced computed
tomography. Invest Radiol 2004; 39: 340–9.
47. Heymann MA, Payne BD, Hoffman JI, and Rudolph AM. Blood flow measure-
ments with radionuclide-labeled particles. Prog Cardiovasc Dis 1977; 20: 55–79.
48. Lykke AW, Cummings R. Increased vascular permeability in the primary
cutaneous allograft response in the rat. Experientia 1969; 25: 1287–8.
49. Purdie TG, Henderson E, Lee TY. Functional CT imaging of angiogenesis in
rabbit VX2 soft tissue. Phys Med Biol 2001; 46: 3161–75.
50. Bland JM, Altman DG. Statistical methods for assessing agreement between
two methods of clinical measurement. Lancet 1986; 1: 307–10.
51. Eliasziw M, Young SL, Woodbury MG, Fryday-Field K. Statistical methodology
for the concurrent assessment of interrater and intrarater reliability: using
goniometric measurements as an example. Phys Ther 1994; 74: 777–88.
52. Goh V, Halligan S, Gartner L, Bassett P, Bartram CI. Quantitative colorectal
cancer perfusion measurement by multidetector-row CT: does greater tumour
coverage improve measurement reproducibility? Br J Radiol 2006; 79: 578–83.
53. Ng QS, Goh V, Klotz E et al. Quantitative assessment of lung cancer perfusion
using MDCT: does measurement reproducibility improve with greater tumor
volume coverage? Am J Roentgenol 2006; 187: 1079–84.
54. Hattori H, Miyoshi T, Okada J et al. Tumor blood flow measured using
dynamic computed tomography. Invest Radiol 1994; 29: 873–6.
9781842143094-Ch03 8/8/07 11:54 AM Page 47
3
Image acquisition and contrast
enhancement protocols for
CT perfusion
Kenneth A Miles
INTRODUCTION AND OVERVIEW:
Most people working in computed tomography (CT) departments are

familiar with image acquisition and contrast enhancement protocols
that aim to delineate the blood vessels supplying a particular organ, for
example CT angiography for the renal or pulmonary arteries. On the
other hand, experience using CT to depict perfusion or other aspects
of vascular physiology within those organs may be less widespread.
Nevertheless, some awareness of vascular physiology is required for the
optimal performance of CT angiography in that the timing of the image
acquisition must be matched to the temporal changes in vascular
enhancement as the contrast medium passes through the circulation.
Many modern CT systems now incorporate an automated method for
timing the image acquisition in accordance with vascular enhancement.
However, prior to this development, it was commonplace to acquire a
rapid sequence of images without table movement during a small ini-
tial test injection of contrast medium. This sequence of images was used
to identify the time of optimal vascular enhancement and so guide the
timing of the main CT angiography study. Conceptually, image acquisi-
tion and contrast protocols for CT perfusion in oncology are similar to
those initial timing sequences, with the temporal changes in contrast
enhancement providing the information necessary for determining
tumor perfusion. However, the focus is not simply upon contrast
enhancement within the major vessels but, more importantly, enhance-
ment within the small vessels inside the tumor itself. In effect, CT perfu-
sion can be considered ‘CT angiography for the tumor microcirculation’.
Just as the timing sequences described above were once performed
in conjunction with CT angiography, CT perfusion data acquisitions can
9781842143094-Ch03 8/8/07 11:54 AM Page 48
be combined with other CT acquisitions for tumor evaluation within

a single examination. Due to the simplicity and speed of acquisition,
the inclusion of CT perfusion adds little time to the examination. For
example, a conventional portal phase post-contrast-enhanced series of
the abdomen and pelvis can be acquired immediately after a CT perfu-
sion acquisition and, in some circumstances, using the same bolus of
contrast medium. CT perfusion can also be performed during selective
hepatic arteriography or superior mesenteric arterial portography to
obtain pure measurements of hepatic arterial and portal perfusion
respectively.1,2 The advent of imaging systems that combine CT with
positron emission tomography (PET) also creates opportunities to
perform CT perfusion as an adjunct to PET imaging of tumor metabo-
lism or receptor activity, etc., without need of the on-site cyclotron
required to obtain perfusion data using PET (Chapter 14).
Figure 3.1 provides an overview of acquisition and contrast-enhance-
ment protocols for CT perfusion. Rather than acquire a series of images
at different anatomical locations across an organ, CT perfusion typically
requires a series of images over time, performed with no movement of
the examination table through the gantry. These ‘raw’ images undergo
computer processing to generate a range of parametric maps of tumor
CT perfusion
Time
Perfusion (m1/min/ml) Blood volume (%) Permeability (µ1/min/ml)
Figure 3.1 Perfusion computed tomography (CT) requires a series of images

performed at a single anatomical location following administration of contrast
medium. The subsequent image processing is analogous to that performed using
CT angiography data
9781842143094-Ch03 8/8/07 11:54 AM Page 49
Image acquisition and contrast enhancement protocols 49
perfusion, blood volume, etc. This processing stage is directly analogous

to deriving maximal intensity projections from CT angiograms, for
example. All CT perfusion protocols require an initial baseline image
with no contrast material. The precise protocol adopted thereafter
depends primarily upon the physiological information required for a
particular clinical scenario and the image processing method used.3
Manufacturers usually offer recommended protocols to be used in
conjunction with their analysis software. This chapter aims to inform
the reader as to the basis behind the design of image acquisition and
contrast enhancement protocols for CT perfusion.
The acquisition and contrast-enhancement parameters to be consid-
ered are (1) the phase of respiration, (2) the overall length of time of
the image series, (3) the number and frequency of images, (4) the
number and thickness of CT slices, (5) the X-ray exposure factors, and
(6) the type, volume, and concentration of contrast medium injected.
Some example protocols are given in Table 3.1. The chosen
protocol will determine the radiation dose received by the patient.
Table 3.1 Example image acquisition and contrast-enhancement protocols for

perfusion computed tomography
Protocol 1 Protocol 2 Protocol 3
Image acquisition
Overall time 60 s 50 s 90 s
Number of images 60 ¥4 25 ¥2 9¥6
Image frequency Every 1 s Every 2 s Every 10s
No. slices ¥thickness Single location: Single location: Multiple spiral:
4 ¥5 mm 2 ¥10 mm 6¥10 mm
Tube voltage 120 kVp 80 kVp 80 kVp
Tube current 50 -100 mAs 100 -200 mAs 100 –200 mAs
Contrast medium
Concentration 370 mg/ml 370 mg/ml 300 mg/ml
Volume 50 ml 40 ml 100 ml
Injection rate 4–7 ml/s 7–10 ml/s Decreasing rate:
4ml/s–2ml/s–1ml/s
Processing
Parameters calculated Perfusion, Perfusion, Standardized
blood volume, blood volume, perfusion value,
mean transit mean transit time permeability,
time, permeability blood volume
Analysis method Deconvolution Single Two compartment
compartment (Patlak)
9781842143094-Ch03 8/8/07 11:54 AM Page 50
Frequently, the desire for image data of higher quality must be bal-
anced against this radiation dose. Although, in the context of oncology,
the radiation exposure associated with CT perfusion is small compared
to the radiotherapy dose that many patients will receive, there remains
a need to limit the radiation burden associated with CT perfusion
studies. A typical radiation dose received from a CT perfusion study of
a limited volume (e.g. 4¥5 mm slices) is between 2 and 10mSv, depend-
ing on the body region, and is therefore comparable to that obtained
from a gamma camera study of tumor perfusion using single photon
emission tomography, for instance. As multidetector CT systems are
developed with larger detector tracks, offering the possibility of single-
location acquisitions over larger tumor volumes, the radiation dose
associated with CT perfusion could potentially increase by a
considerable amount. It should also be remembered that the aim of
CT perfusion is to produce physiological data rather than display
anatomical structure, and, thus, spatial resolution, although important,
is not the primary consideration.
IMAGE ACQUISITION
Phase of respiration
Respiratory motion can result in significant misregistration of images
acquired during the dynamic image sequence for CT perfusion in the
chest and abdomen. Misregistration leads in turn to errors in perfusion
values and artifacts on parametric images. CT perfusion protocols for
chest and abdomen are usually acquired with either suspended respi-
ration or quiet breathing. Suspended respiration is unlikely to be appro-
priate for image series longer than 45–60 seconds, unless acquisition
pauses are created to allow the patient to take a breath, as might be the
case for multiple spiral protocols. Many patients who have suspended
their breathing will slowly exhale during the acquisition, resulting in
motion artifacts. Protocols that adopt quiet respiration can result in
high-quality perfusion images, but the patient must be warned to avoid
the temptation to take a deep breath when experiencing the ‘hot flush’
commonly associated with a rapid bolus of contrast medium. Quiet res-
piration will also be appropriate when co-registering CT perfusion data
with PET images, acquired using an integrated PET–CT system, as
suspended respiration is not possible for the PET images, which take
several minutes. Respiratory gating of CT images has the potential
to reduce CT perfusion motion artifacts in the future.
9781842143094-Ch03 8/8/07 11:54 AM Page 51
Overall length of time of the image series

The overall length of the image series governs which physiological
parameters can be measured. Tumor blood flow, blood volume, and
mean transit time (MTT) can be assessed during the first pass of contrast
material through the vascular system, at which time the contrast mate-
rial is predominantly intravascular. The length of the first pass depends
upon an individual’s cardiac output and circulating blood volume, but
typically comprises the first 45–60 seconds immediately after intra-
venous injection (e.g. Table 3.1, protocols 1 and 2). A longer series is
required for the assessment of tumor vascular permeability (e.g. Table 3.1,
protocol 3). This additional time is needed to allow sufficient contrast
material to pass into the extravascular space for reliable measurements.
Using the deconvolution technique to assess colorectal cancer, Goh
et al. showed that acquisition times of 45 s resulted in significantly lower
values for vascular permeability than acquisitions of 65 s or more,
whereas measurements of blood flow, blood volume, and MTT were
consistent across all acquisition times between 45 s and 130 s.4
The number and frequency of images

The number and frequency of images obtained will be a balance
between the quality of the perfusion data and the radiation dose for the
patient. A high image frequency will be required for reliable measure-
ments from first pass data. Although modern CT systems can acquire
images with a frequency of faster than one image every second, studies
of cerebral perfusion have shown that image frequency can be reduced
to every 2–3 seconds without adversely affecting the perfusion values
obtained (e.g. Table 3.1, protocol 2).5 To assess vascular permeability,
images can be acquired less frequently (every 5–10 s), as changes in
contrast enhancement become less rapid (e.g. Table 3.1, protocol 3).
The number and thickness of CT slices

When focusing on the first pass of contrast medium, the volume of
tumor included in a CT perfusion study is usually constrained in the
craniocaudal (z axis) direction by the width of the CT detector, typically
2–4cm for current multidetector (MDCT) systems. This limitation is due
to the need to acquire images at a frequency close to the time taken for
the CT system to rotate the X-ray tube around the patient (typically 0.5–1 s)
which does not allow sufficient time for tabletop movement during the
study. The detector can be divided so that a number of slices are
obtained simultaneously, e.g. four slices 5 mm thick. Although slices
9781842143094-Ch03 8/8/07 11:54 AM Page 52
as thin as 0.5 mm can be obtained, reducing slice thickness increases

image noise, which can only be compensated effectively by increasing
the radiation exposure.
Moving the imaging table through the gantry during acquisition
allows greater volumes of tissue to be examined. A table-toggling
approach, in which table movement oscillates during a continuous
acquisition, obviates the delay that would otherwise be required
to return the table to its starting position.6 The greater is the extent of
table movement, the lower is the frequency of image acquisition obtain-
able at a particular slice location. Using a 1-second tube rotation, table-
toggling a distance that is twice the width of the detector tract
(i.e. approximately 4–8 cm on current MDCT systems) will keep the
cycle time for data acquisition at each slice location to within 2 seconds,
thereby allowing reliable perfusion measurements.
A further increase in the tumor volume assessed can be achieved by
using repeated spiral acquisitions, but the greater coverage is off-set
by the necessary increase in cycle time. Such protocols are suitable
for the assessment of tumor vascular permeability and blood volume
(e.g. Table 3.1, protocol 3). Repeated spiral acquisitions are also
frequently used for measurements of peak tumor enhancement, for
instance in the characterization of pulmonary nodules7 (see Chapter 7).
With suitable calibration of the CT system, such measurements of peak
enhancement can also be converted to the standardized perfusion value
(SPV).8 Due to the reduced image frequency, there is potential for such
protocols to miss the time of peak tumor enhancement. However, this
error may be small for tumors exhibiting increased vascular permeabil-
ity, because the rapid passage of contrast material into the extravascu-
lar space will reduce the washout of contrast material following the first
pass.
A technique that uses a single spiral (helical) volume acquisition
close to the time of peak cerebral enhancement during the first pass of
contrast medium has been proposed for the measurement of cerebral
volume over the whole brain.9 This method could potentially be
applied to tumor imaging in other body regions, to increase the volume
of tumor assessed without a major increase in radiation burden. Correct
timing could be achieved by using a single-location timing sequence
following a small test-bolus of contrast medium to determine the time
of peak tissue enhancement (analogous to the timing sequences often
used during CT angiography to determine the time of peak vascular
enhancement). Data from this timing sequence could also be processed
to obtain absolute perfusion data with the chosen slice level (Figure 3.2).
Although such a protocol would allow measurements of tumor blood
9781842143094-Ch03 8/8/07 11:54 AM Page 53
1. Low mA pre-contrast spiral acquisition
2. Single-location time sequence (low mA, small volume of contrast)
3. Spiral acquisition at time of peak tumor enhancement:

superimpose peak enhancement image (color) on pre-contrast CT
Figure 3.2 CT perfusion technique that combines single-level measurements of

perfusion, blood volume, and transit time with a volumetric assessment of tumor
blood volume and standardized perfusion value
volume and SPV, assessment of MTT and vascular permeability would

not be possible with this technique.
Greater anatomical coverage can allow compensation for respiratory
misregistration by retrospective selection of sets of images from the whole
volume, in which the tumor is displayed at the same craniocaudal loca-
tion. This technique has been successfully applied to CT perfusion of
the liver.10 Portal perfusion measurements without respiratory compen-
sation were significantly higher than when compensation was applied.
An increase in the extent of tumor assessed by CT perfusion also has
the potential to reduce measurement errors resulting from any hetero-
geneity of the tumor vasculature. However, the level of benefit may be
dependent upon tumor type and analysis method. Using deconvolution
analysis in colorectal cancer, Goh et al. found no improvement in meas-
urement reproducibility when increasing z-axis coverage from 5 mm to
20 mm,11 whereas for Patlak analysis of lung tumors, Ng et al. found that
reproducibility was improved by increasing z-axis coverage from 10 mm
to 40 mm.12
X-ray exposure factors

Tube voltage (kVp), tube current (mA), and exposure time are important
aspects of the acquisition protocol. Higher values for each of these
9781842143094-Ch03 8/8/07 11:54 AM Page 54
parameters reduce image noise but increase radiation dose for the
patient. Typical exposure times for current CT scanners are 1 second or
less. Selecting a smoother reconstruction filter (e.g. soft-tissue) will
reduce image noise, but will also reduce spatial resolution.
The tube voltage also governs the amount of attenuation produced
by a given iodine concentration of contrast medium (Figure 3.3).
A greater increase in attenuation is produced at tube voltages lower
than that typically used for diagnostic studies (e.g. 80–100kVp instead
of 120–140 kVp). However, a lower tube voltage heightens image noise
and beam hardening effects. The precise relationship between attenua-
tion change and iodine concentration varies between CT systems, with
aging of the X-ray tube, and for different slice positions across the track
of a multidetector CT system.13 The likely explanation for these varia-
tions in iodine response is differences in the X-ray spectra generated by
a particular X-ray tube. The X-ray spectrum will be modified by beam
hardening within the anode, which will not only change as the anode
wears over time but also vary across the X-ray beam (anode heal
effect). Because algorithms for calculating perfusion from CT typically
divide tissue enhancement measures by vascular enhancement, the
effects of variable iodine response will cancel out for a given perfusion
measurement, provided that the tissue and vascular enhancement data
are obtained from the same portion of the detector track.
The different analysis methods used in processing CT perfusion data
(see Chapter 2) are more or less sensitive to image noise, and thus
the X-ray exposure should be matched to the processing algorithm.
800
600
Attenuation (HU)
400
80 kVp
200
100 kVp
120 kVp
140 kVp
0
0 5 10 15 20
Iodine concentration (mg/ml)
Figure 3.3 Attenuation values associated with different concentrations of contrast

medium measured in a phantom using tube voltages between 80 and 140 kVp.
9781842143094-Ch03 8/8/07 11:54 AM Page 55
Deconvolution analysis uses all the images within an acquisition

sequence, and is therefore affected less by noise in individual images.
Hence, a lower tube current can be selected, allowing a higher image
frequency (e.g. protocol 1, Table 3.1). However, single-compartment
analysis depends upon accurate attenuation measurements within criti-
cal images in the sequence, and too much image noise can result in
overestimation of tissue enhancement rates with miscalculation of per-
fusion values. Thus, time–attenuation data from protocols that adopt a
higher tube current (with a lower image frequency to minimize the
radiation dose) are appropriate for compartmental analysis methods
(e.g. protocols 2 and 3, Table 3.1). Such data sets can also be success-
fully processed using deconvolution analysis (Griffiths, personal
communication, 2001).
CONTRAST ENHANCEMENT
Type of contrast medium

Non-ionic contrast medium is preferred to an ionic preparation due to
the lower effects of flushing and nausea. Such side-effects increase the
chance of the patient moving during the image sequence, resulting in
errors due to misregistration between sequential images. The perform-
ances of the non-ionic contrast media for CT perfusion are broadly
similar, although measurements of vascular permeability may vary
slightly for different contrast media depending on the molecular weight.
Volume and concentration of contrast medium

Regardless of the analysis method, first-pass studies of tumor perfusion
will be improved by maximizing contrast enhancement in blood vessels
and tissues to produce better signal-to-noise ratios for time–attenuation
data. Maximal enhancement can be achieved by a rapid injection rate
(at least 4ml/s) and by using contrast media with a high iodine concen-
tration (i.e. 350–400mg iodine/ml, Figure 3.4).
When using compartmental analysis, perfusion measurements are
affected by the width of the bolus of contrast medium in the vascular
system (Figure 3.4). For this analysis method, the injection rate should
be at least 7ml/s, and the volume of contrast medium no greater than
50ml (protocol 2, Table 3.1). Some authors suggest injection rates of up
to 10 ml/s, but injections above this rate are unlikely to improve the
bolus further on account of buffering in the venous system between the
9781842143094-Ch03 8/8/07 11:54 AM Page 56
1. Narrow vascular bolus

Artery
2. Maximized tissue
HU
contrast enhancement
Tissue
Time
Figure 3.4 Time–attenuation curves for CT perfusion are improved by use of

high-concentration contrast media
site of injection and the heart. A narrow vascular bolus is needed

because the validity of compartmental analysis requires that no contrast
medium should leave the tumor region of interest by the time of meas-
urement. The magnitude of error resulting from failure to meet this
requirement will depend upon the minimum transit time (minTT) and
mean transit time (MTT) for passage of contrast material through the
brain tissue. Table 3.2 demonstrates the outcome of a computer simu-
lation to assess the magnitude of error for a range of mean and mini-
mum tissue transit times. The results are based on typical arterial
time–attenuation data acquired during a CT perfusion study using a
50ml bolus of contrast medium administered with an injection rate of
7ml/s. It can be seen that, provided the minTT is at least 4 seconds, the
error is small (i.e. <5%), even when the MTT is as short as 6 seconds.
Reported MTT values for tumors range from approximately 4.5s in
squamous cell carcinoma of the oropharynx,14 through 8 seconds for
colorectal cancer,4,15 to 20 s in cervical cancer.16
Table 3.2 Effect of tissue transit times on underestimation of perfusion (expressed as

percentage error) determined by compartmental analysis for a typical arterial input from
50 ml of contrast medium injected at 7 ml/s
MTT (s) MinTT = 2s (%) MinTT = 4s (%) MinTT = 6s
6 9.51 3.29 0.03

8 6.92 1.86 0.03
10 5.44 1.30 0.03
12 4.47 1.00 0.03
14 3.80 0.82 0.03
16 3.31 0.69 0.03
MTT, mean transit time; minTT, minimum transit time

9781842143094-Ch03 8/8/07 11:54 AM Page 57
Measurements of tumor vascular permeability are less affected by

bolus shape. In fact, when using Patlak analysis (see Chapter 2), there
is a theoretical advantage from maintaining a more constant intravascu-
lar concentration of contrast material by using a decreasing injection
rate (Table 3.1, protocol 3).12
CHOICE OF PROTOCOL
In certain organs, the anticipated physiological differences between

tumor and normal tissues may influence the choice of protocol. For
example, the longer acquisitions required for vascular permeability
measurement may be particularly suitable for the study of brain tumors,
because cerebral glioma is typically associated with permeability values
much higher than that of the almost impermeable intact blood–brain
barrier, whilst tumor perfusion and blood volume values may closely
approximate those of normal cerebral cortex. When assessing tumor
response to antiangiogenesis agents, the mechanism of action of a
specific agent may favor measurement of one particular physiological
parameter, and hence favor a particular acquisition protocol. One such
example is provided by agents that block vascular endothelial growth
factor (VEGF), which may be more likely to produce effects upon
vascular permeability. On the other hand, image misregistration from
respiratory or other motion is more likely to affect longer acquisition
protocols, which may therefore be more problematic in the chest and
abdomen. In general, consideration should be given to matching the
CT perfusion protocol used to the particular application intended.
COMPARING RESULTS OBTAINED USING DIFFERENT

PROTOCOLS
The lack of standardization of acquisition and analysis protocols for

CT perfusion creates a constraint on the ability to compare the results
of different studies. These difficulties are compounded by variability in
the response of different CT systems to iodine, and even variability in
the same system with time.13 Much would be gained if there were a
widely adopted quality-assurance process that calibrated CT systems for
iodine response. This section proposes ‘rules-of-thumb’ to enable com-
parison between different CT perfusion techniques. However, in the
absence of widespread CT system calibration, these comparison methods
can only offer approximations.
9781842143094-Ch03 8/8/07 11:54 AM Page 58
Tumor enhancement
Tumor enhancement values are more readily compared when expressed
as a ratio of enhancement to injected dose of contrast material (e.g. HU/g
iodine). However, it is important to compare other factors such as rate
of injection, time of image acquisition, and, most important, the tube
voltage used. A reduction in tube voltage from 120kVp to 80kVp can
increase the sensitivity of a CT system to iodine by as much as 50%.13
For example, let us compare the technique for assessing the likelihood
of malignancy within pulmonary nodules proposed by Swensen et al.7
to the study by Shim et al. who used enhancement of T1 lung cancers to
predict the likelihood of nodal metastases.17 Swensen et al. adopted an
enhancement threshold of 15 HU, whereas Shim et al. used 60 HU. Both
studies used a tube voltage of 120kVp. In the Swensen study, a typical
70-kg man would have received 29.4 g of iodine injected over 49 s, and
thus the enhancement threshold could be expressed as 0.51HU/g iodine.
Shim et al. used a 60-HU enhancement threshold following an injection of
36g of iodine administered over a similar time (40 s), giving a corrected
threshold of 1.67HU/g iodine. It can be seen that, although the enhancement
threshold associated with an increased likelihood of nodal metastases is
four times that needed to diagnose malignancy, this difference is nearer
three times when enhancement is corrected for the dose of iodine
injected.
Comparing tumor perfusion and enhancement

Based on the Fick principle, tumor perfusion can be determined from
the peak enhancement, the dose of contrast material given, the cardiac
output, and a CT calibration factor that relates iodine concentration
to attenuation changes measured in Hounsfield units (see Box 3.1).8
Box 3.1 Calculation of perfusion from peak enhancement
Perfusion (ml/min/100 ml) =

Peak enhancementt (HU)
× 10 0
Area under aortic curve (HU·min)
Cardiac output (ml/min) =
Dose of contrastt medium (g) × Calibration factor (HU/g/ml)
Area under aortic curve (HU⋅min)
Perfusion (ml/min/100 ml) =
Peak enhancement (HU) Cardiac output (ml/min)
× ×100
0
Dosse of contrast medium (g) Calibration factor (HU/g/ml)
9781842143094-Ch03 8/8/07 11:54 AM Page 59
Thus, by using typical calibration factors and making some basic

assumptions concerning the cardiac output, it is possible to derive a
conversion factor that approximates tumor enhancement to perfusion.
Let us consider again the enhancement thresholds used by Swensen
et al.7 and Shim et al.17 for the evaluation of lung lesions, as discussed
above. Assuming a cardiac output of 5000ml/min and a calibration
factor for 120 kVp in air of 27 400 HU/g/ml (based on the median value
reported by Miles et al.13), the conversion factor that relates perfusion
to enhancement per gram of iodine is 5000/27 400¥100 = 18. Thus, the
approximate perfusion threshold for diagnosis of malignancy would be
9.2 ml/min/100 ml, and for increased likelihood of nodal metastasis
30 ml/min/100 ml. These values are compatible with the range of lung
tumor perfusion reported by Kiessling et al.18
SUMMARY
The challenge for CT perfusion acquisition protocols is to image a

larger volume of tissue whilst keeping radiation exposure within
reasonable limits. Many of the variations among protocols aim to
minimize dose rather than improve the accuracy of perfusion measure-
ments. The key factors for optimizing the accuracy of measurements of
tumor perfusion are a low tube current (e.g. 80 kVp), a short bolus
of high-concentration contrast medium, and a frequency of acquisition
of at least one image every 2–3 seconds. Overall CT perfusion is a
simple and robust technique that is readily incorporated into existing
CT protocols.
REFERENCES
1. Komemushi A, Tanigawa N, Kojima H, Kariya S, Sawada S. CT perfusion of

the liver during selective hepatic arteriography: pure arterial blood perfusion
of liver tumor and parenchyma. Radiat Med 2003; 21: 246–51.
2. Kojima H, Tanigawa N, Komemushi A, Kariya S, Sawada S. Computed tomog-
raphy perfusion of the liver: assessment of pure portal blood flow studied with
CT perfusion during superior mesenteric arterial portography. Acta Radiol
2004; 45: 709–15.
3. Miles KA. Perfusion CT for the assessment of tumour vascularity: which pro-
tocol? Br J Radiol 2003; 76 (Spec No 1): S36–42.
4. Goh V, Halligan S, Hugill JA, Gartner L, Bartram CI. Quantitative colorectal
cancer perfusion measurement using dynamic contrast-enhanced multidetec-
tor-row computed tomography: effect of acquisition time and implications for
protocols. J Comput Assist Tomogr 2005; 29: 59–63.
9781842143094-Ch03 8/8/07 11:54 AM Page 60
5. Wintermark M, Smith WS, Ko NU et al. Dynamic perfusion CT: optimizing the

temporal resolution and contrast volume for calculation of perfusion CT
parameters in stroke patients. Am J Neuroradiol 2004; 25: 720–9
6. Roberts HC, Roberts TP, Smith WS et al. Multisection dynamic CT perfusion for
acute cerebral ischemia: the ‘toggling-table’ technique. Am J Neuroradiol 2001;
22(6): 1077–80.
7. Swensen SJ, Viggiano RW, Midthun DE et al. Lung nodule enhancement at CT:
multicenter study. Radiology 2000; 214: 73–80.
8. Miles KA, Griffiths MR, Fuentes MA. Standardized perfusion value: universal
CT contrast enhancement scale that correlates with FDG PET in lung nodules.
Radiology 2001; 220: 548–53.
9. Hunter GJ, Hamberg LM, Ponzo JA et al. Assessment of cerebral perfusion and
arterial anatomy in hyperacute stroke with three-dimensional functional CT:
early clinical results. Am J Neuroradiol 1998; 19: 29–37.
10. Nakashige A, Horiguchi J, Tamura A et al. Quantitative measurement of
hepatic portal perfusion by multidetector row CT with compensation for
respiratory misregistration. Br J Radiol 2004; 77: 728–34.
11. Goh V, Halligan S, Gartner L, Bassett P, Bartram CI. Quantitative colorectal
cancer perfusion measurement by multidetector-row CT: does greater tumour
12. Ng QS, Goh V, Klotz E et al. Quantitative assessment of lung cancer perfusion
using MDCT: does measurement reproducibility improve with greater tumor
volume coverage? Am J Roentgenol 2006; 187: 1079–84.
13. Miles KA, Young H, Chica SL, Esser PD. Quantitative contrast-enhanced com-
puted tomography: is there a need for system calibration? Eur Radiol 2007; 17:
919–26.
14. Gandhi D, Chepeha DB, Miller T et al. Correlation between initial and early
follow-up CT perfusion parameters with endoscopic tumor response in
patients with advanced squamous cell carcinomas of the oropharynx treated
with organ-preservation therapy. Am J Neuroradiol 2006; 27: 101–6.
15. Sahani DV, Kalva SP, Hamberg LM et al. Assessing tumor perfusion and treat-
ment response in rectal cancer with multisection CT: initial observations.
Radiology 2005; 234: 785–92.
16. Haider MA, Milosevic M, Fyles A et al. Assessment of the tumor microenviron-
ment in cervix cancer using dynamic contrast enhanced CT, interstitial fluid
pressure and oxygen measurements. Int J Radiat Oncol Biol Phys 2005; 62:
1100–7.
17. Shim SS, Lee KS, Chung MJ et al. Do hemodynamic studies of stage T1 lung
cancer enable the prediction of hilar or mediastinal nodal metastasis?
Am J Roentgenol 2006; 186: 981–8.
18. Kiessling F, Boese J et al. Perfusion CT in patients with advanced bronchial
carcinomas: a novel chance for characterization and treatment monitoring? Eur
Radiol 2004; 14: 1226–33.
9781842143094-Ch04 8/8/07 11:55 AM Page 61
4
Image processing
Ting-Yim Lee, Xiaogang Chen, and Kenneth A Miles
INTRODUCTION
The methods for deriving tumor perfusion data from dynamic contrast-
enhanced computed tomography (CT) images outlined in the previous
chapters can be applied to regions of interest (ROI) constructed on the
basis of anatomical features visible on conventional CT images.
However, abnormalities of tumor perfusion may not be readily appar-
ent on conventional CT images, and therefore the ability to place ROIs
over areas of physiological importance is severely limited. If perfusion
values are calculated using time–attenuation data derived from individ-
ual pixels, it is possible to generate a parametric map depicting blood
flow throughout the whole CT slice chosen for the dynamic study.
The spatial resolution of these CT perfusion images is identical to that
of conventional CT, and thus superior to other techniques such as
positron emission tomography or even magnetic resonance imaging.
Indeed, in addition to images of tumor blood flow, a whole range of
parametric maps can be produced in this way (Table 4.1). Figure 4.1
shows the appearance time (Figure 4.1a), blood flow (Figure 4.1b),
blood volume (Figure 4.1c), and capillary permeability surface area
product (Figure 4.1d) of a brain tumor in the right cerebral hemisphere.
The tumor (arrow in Figure 4.1d) is clearly delineated in the blood flow
(F), blood volume (Vb), and capillary permeability surface area product
(PS) parametric maps as having a hypervascular rim surrounding a
hypovascular core, with high blood flow, blood volume, and capillary
permeability surface area product in the rim but low values in the core.
There are also two distant metastases (arrowheads in Figure 4.1d)
which are visible as focal regions of increased PS and blood flow. In
the normal brain, there is clear differentiation between gray and white
matter in both blood flow and blood volume. In the appearance time
map, the distinction between gray and white matter is less, suggesting
9781842143094-Ch04 8/8/07 11:55 AM Page 62
Table 4.1 Summary of computed tomography (CT) tumor perfusion parameters

commonly displayed as parametric maps
Operational equations
Parameter Method of derivation for derviation*
Tumor blood flow (F) Maximal slope method Equation (7a)

Distributed parameter model Equations (5a) and (23)
Tumor blood Two-compartment model Equations (5a) and (21) or (22)
volume (Vb) Distributed parameter model Equations (5a) and (23)
Mean transit time (Tm) Distributed parameter model Equations (5a) and (23)
Maximum Peak pixel attenuation – —
enhancement baseline attenuation
Arrival time of contrast Two-compartment model Equations (5a) and (21) or (22)
with respect to Distributed parameter model Equations (5a) and (23)
arterial input (T0)
Perfusion-weighted Sum of the raw CT perfusion —
map (proportional images over the first
to perfused blood contrast circulation peak.
volume)
Capillary permeability Distributed parameter model Equations (5a) and (23)
surface area
product (PS)
Flow extraction Two-compartment model Equations (5a) and (21) or (22)
product (FE)† Distributed parameter model Equations (5a) and (23)
Hepatic arterial Two-compartment model Equations (5a), (21), and (24)
fraction (α) Distributed parameter model Equations (5a), (23), and (24)
*Refer to Chapter 2 for discussion of the equations listed. For liver studies, equation (24) has to be
included.
†
As discussed in Chapter 2, for F >> PS, FE ≈ PS; and for F << PS, FE ≈ F.
that the arrival times for gray and white matter are more similar than
their blood flows and volumes.
In some studies, the visualization of areas of physiological difference
can be obscured by image noise. Examples include the higher perfusion
values within cerebral cortex as compared to white matter (Figure 4.2a),
and tumors with perfusion values close to that of the surrounding tissue
(Figure 4.3a). Performing a spatial smoothing process can reduce the
impact of noise by averaging out the individual pixel values over a
larger area, and so displays physiological differences more clearly
(Figures 4.2b and 4.3b). It is usual for a weighted average to be
adopted, such that the pixel values immediately adjacent to the pixel
under consideration are weighted more heavily than pixels further
away. An example is 3 × 3 binomial smoothing whose weighting
factors are summarized in Table 4.2. In this example, the central pixel
9781842143094-Ch04 8/8/07 11:55 AM Page 63
Image processing 63
a b
5 90
0 0
c d
5 30
0 0
Figure 4.1 Functional maps from a computed tomography (CT) tumor perfu-
sion study in a patient with a glioblastoma multiforme in the right cerebral
hemisphere: (a) appearance time (T0) map, pixel values in units of seconds;
(b) blood flow (F) map, pixel values in units of ml/min/100 g; (c) blood volume
(Vb) map, pixel values in units of ml/100 g; (d) capillary permeability surface
area product (PS) map, pixel values in units of ml/min/100 g. The arrow points
to the main tumor whilst the arrowheads indicate two distant metastases
a b
90 90
0 0
Figure 4.2 Blood flow maps of the same patient as in Figure 4.1 without
(a) and with (b) binomial spatial smoothing. Note that the higher perfusion
within gray matter structures is more readily apparent on the smoothed image
9781842143094-Ch04 8/8/07 11:55 AM Page 64
a b
Figure 4.3 Hepatic arterial perfusion maps displayed without (a) and with (b)
binomial spatial smoothing. Note that the higher perfusion within the hepatic
metastasis (arrow) is more readily apparent on the smoothed image
receives greatest weight, followed by orthogonally adjacent pixels, with

diagonally adjacent pixels weighted least.
DETERMINATION OF AVERAGE MEAN TRANSIT TIME

OF A REGION OF INTEREST
As discussed above, using a region of interest (ROI) to find the mean

value of tumor perfusion within the region is a convenient method of
reducing the perfusion map to a single number for further data analysis.
The average perfusion, F, in a ROI is defined as:
1 N
F= ∑F
N i=1 i
(1a)
where N is the number of pixels in the ROI and Fi is the perfusion in

the ith pixel. A similar calculation can be used to find the average blood
volume, Vb , within a ROI:
1 N
V= ∑V
N i=1 b,i
(1b)
Table 4.2 Weighting factors for individual pixels for 3 × 3 binomial spatial smoothing
1 2 1
2 4 2
1 2 1
9781842143094-Ch04 8/8/07 11:55 AM Page 65
Image processing 65
where Vb,i is the blood volume in the ith pixel. It may therefore be sur-
prising at first glance to find out that the average mean transit time, Tm ,
cannot be calculated, similar to equation (1a) and (1b), as:
1 N
Tm = ∑T
N i=1 m,i
(2)
where Tm,i is the mean transit time through the ith pixel. Consider a ROI
consisting of two pixels. In one pixel blood flow is zero, but blood
volume can be either zero or non-zero; by definition the mean transit
time in this pixel is zero. In the second pixel, blood flow, blood
volume, and mean transit time are all finite. In this case, the mean tran-
sit time of the ROI should be equal to the finite mean transit time of the
2nd pixel, Tm,2; however, according to equation (2), it is equal to one-
half of Tm,2 instead, which is incorrect. The proper definition of the
average mean transit time of a ROI should be:
N N
Fi
Tm = ∑ T where FT = ∑ Fi (3)
i=1 FT m,i i=1
or, the average mean transit time of a ROI is the flow weighted
average of mean transit time of each pixel within the ROI. Since
Fi ◊ Tm,I = Vb,I (4)
because of the central volume principle, equation (3) simplifies into:
1 N
1 N Vb,T Vb
Tm =
FT
∑F ⋅T
i=1
i m,i
=
FT
∑V
i=1
b,i
=
FT
=
F
(5)
Equation (4) states that the central volume principle applies to average
values of blood flow, blood volume, and mean transit time from a ROI
in the same way as it applies to values from single pixels, provided that
the average mean transit time of a ROI is defined as in equation (3).
IMAGE PROCESSING
ROIs placed manually are more or less operator-dependent, and thus

potentially contribute to interobserver variations in derived perfusion
values.1 It is therefore desirable to incorporate image processing
techniques that automatically identify a range of tissue and vascular
structures, a process known as ‘image segmentation’.
9781842143094-Ch04 8/8/07 11:55 AM Page 66
Segmentation of specific organs and tumor

CT perfusion images of many organs are improved by eliminating areas
of fat, calcification, and bone from the image. This can be achieved by
removing any pixels with initial attenuation values above and below a
predefined range. For example, cerebral tissues on CT scans without
contrast media typically have attenuation values between 0 HU and 80 HU,
and thus non-cerebral tissues are readily removed from CT perfusion
images by the use of thresholds based on these attenuation measure-
ments (Figure 4.4a–c). CT perfusion images of some tumors can be sim-
ilarly improved by the use of attenuation thresholds for segmentation;
for example, lung tumors can be separated from adjacent lung tissue
(Figure 4.5a–c).
a b
c d
Figure 4.4 (a) Contrast-enhanced CT and maximum enhancement images

(b) without segmentation, (c) with elimination of non-cerebral tissues, and (d)
with elimination of non-cerebral tissues and vascular structures. Non-cerebral
tissues were removed by excluding pixels with attenuation values before admin-
istration of contrast medium of less than 0 HU and more than 60 HU. Vascular
structures were eliminated by removing pixels with peak enhancement more
than 10% of the peak enhancement within the sagittal sinus
9781842143094-Ch04 8/8/07 11:55 AM Page 67
Image processing 67
a b
c d
Figure 4.5 (a) Contrast-enhanced CT and perfusion images (b) without seg-
mentation, (c) with elimination of non-tumor tissues, and (d) with elimination
of non-tumor tissues and vascular structures. Non-tumor tissues were removed
by excluding pixels with attenuation values before administration of contrast
medium of less than 0 HU and more than 100 HU. Vascular structures were
eliminated by removing pixels with peak enhancement more than 10% of the
peak enhancement within the aorta
Segmentation of vascular structures

It is advantageous to automatically identify vascular structures during
processing of CT perfusion images for two reasons: (1) to create greater
consistency in generation of the vascular time–attenuation curves
required as input functions for the perfusion analysis, and (2) to
exclude from CT perfusion images peripheral blood vessels that may
falsely appear as areas of high perfusion.
Automated algorithms that determine vascular input ROIs often iden-
tify a subset of pixels within a larger ROI placed by the operator to
encompass a candidate blood vessel. Possible selection criteria for
these pixels include the timing and level of peak contrast enhancement
achieved. For example, one algorithm calculates the maximum
enhancement of a group of five pixels within a region, and selects those
pixels within 10% of this enhancement and with a time of maximum
enhancement within 10 seconds of those pixels.
Criteria for elimination of peripheral blood vessels include enhance-
ment values and perfusion values (Figure 4.4d). For example, for CT
9781842143094-Ch04 8/8/07 11:55 AM Page 68
perfusion imaging of the brain, Kudo et al.2 eliminated pixels with per-
fusion values above 8 ml min−1100 g−1, whilst Sase et al.3 defined crite-
ria based on ratios of peak enhancement values within pixels to the
peak enhancement within the superior sagittal sinus. Similar processes
can be applied to non-cerebral tumors located close to large or
medium-sized blood vessels (Figure 4.5d). Elimination of vascular
pixels in this way improves the accuracy of perfusion measurements.2
Segmentation of different tissue types

In a CT tumor perfusion study, the same tissue slice is scanned multi-
ple times during the passage of a contrast bolus; a parametric image
can be created by adding (averaging) all images of the slice together.
Since the value of each pixel in the summed (averaged) image is related
to the area under the tissue residue curve of the pixel, the summed
(averaged) image depicts the distribution of perfused blood volume in
the slice,4 and is also called the perfusion-weighted map. Figure 4.6a is
the perfusion-weighted map of the same brain tumor patient as in
Figure 4.1. It shows a clear distinction between gray and white matter.
The separation between gray and white matter is a result of two factors:
the inherent attenuation difference and perfused blood volume differ-
ence between the two tissue types. Figure 4.6b shows the automatic
segmentation of subcortical white matter from cortical gray matter in
normal tissue, as well as enhancing from non-enhancing regions in the
tumor. This segmentation has been achieved by applying thresholds to
the perfusion-weighted map inside ROIs drawn roughly around the
whole brain, the fissure between the cerebral hemispheres, and the
primary tumor and two metastastic deposits as identified by the PS map
in Figure 4.1d. An advantage of using the perfusion-weighted map is
that segmentation of the different tissue types is performed with the
original high spatial resolution of the raw CT perfusion image and
enhanced signal-to-noise over the raw images due to the summation
(averaging) operation. In addition, the perfusion-weighted map is
perfectly aligned with blood flow and other functional maps because
they are derived from the same set of image data. Thus, the pixel maps
of gray and white matter in normal tissue and of enhancing regions
in primary tumor and metastases identified in Figure 4.6b can be
directly applied to the functional maps, for example the blood flow
map (Figure 4.6c), the blood volume map (Figure 4.6d), and the PS
map (not shown), to determine the average blood flow, blood volume,
and PS in the different tissue types, as shown in Tables 4.3, 4.4, and 4.5
respectively. Table 4.6 compares average mean transit time in the
9781842143094-Ch04 8/8/07 11:55 AM Page 69
Image processing 69
a b
720 720
0 0
c d
90 5
0 0
Figure 4.6 (a) Perfusion-weighted image of the same patient as in Figure 4.1
with five regions of interest (ROIs) drawn roughly to outline the whole brain (1),
the intercerebral fissure (2), the primary tumor (3), and two metastases (4 and 5);
(b) thresholding to highlight gray matter () and white matter () in normal
tissue and enhancing () and non-enhancing regions () in tumor; (c) the
pixel masks of gray and white matter as well as enhancing and non-enhancing
tumor regions applied to the blood flow map to determine the average blood
flow in gray matter or enhancing tumor regions () and white matter or non-
enhancing tumor regions (); (d) the pixel masks of gray and white matter
as well as enhancing and non-enhancing tumor regions applied to the
blood volume map to determine the average blood volume in gray matter or
enhancing tumor regions () and white matter or non-enhancing tumor
regions ()
different tissue types of Tables 4.3–4.5, calculated by simple averaging

(equation (2)) and flow weighted averaging (equation (5)). As dis-
cussed above, the average mean transit times for ROIs should be calcu-
lated using flow weighted averaging. Table 4.6 shows that simple
averaging using equation (2) gives rise to average mean transit times
that differ substantially from the correct flow weighted averages.
9781842143094-Ch04 8/8/07 11:55 AM Page 70
Table 4.3 Average blood flow in the different tissue types identified by automatic
thresholding in the regions of interest (ROIs) shown in Figure 4.6a
Blood flow ml/min/100g

Tissue type Average SD Pixel count
Lateral normal cortical gray matter 28.7 18.3 20145

(ROI 1–ROI 2)
Lateral normal subcortical white matter 17.2 10.3 21518
(ROI 1–ROI 2)
Medial normal cortical gray matter 34.3 21.9 9828
(ROI 2)
Medial normal subcortical white matter 17.0 10.2 7337
(ROI 2)
Enhancing primary tumor (ROI 3) 60.3 26.4 1599
Non-enhancing primary tumor (ROI 3) 18.5 15.6 1808
Enhancing metastasis (ROI 4) 59.3 18.0 348
Non-enhancing metastasis (ROI 4) 35.1 10.0 108
CONCLUSION
Pixel-by-pixel analysis of the temporal changes in tumor enhancement

enables a range of parametric maps to be produced. The utility of these
maps is enhanced by image processing techniques that correct for
Table 4.4 Average blood volume in the different tissue types identified by automatic
thresholding in the ROIs shown in Figure 4.6a
Blood volume ml/100g


(ROI 1–ROI 2)
(ROI 1–ROI 2)
(ROI 2)
(ROI 2)
9781842143094-Ch04 8/8/07 11:55 AM Page 71
Image processing 71
Table 4.5 Average capillary permeability surface area product (PS) in the different tissue
types identified by automatic thresholding in the ROIs shown in Figure 4.6a
PS ml/min/100 g

(ROI 1–ROI 2)
(ROI 1–ROI 2)
(ROI 2)
(ROI 2)
noise, segment-specific organs, tumors, and vascular structures, and

automatically identify different tissue types. An important caveat is that
whereas average blood flow and volume for a ROI can be calculated
with simple averaging, average mean transit time has to be properly
calculated with flow weighted averaging.
Table 4.6 Comparison of average mean transit time (equation (2)) and flow weighted
average mean transit time (equation (5)) in the different tissue types identified by
automatic thresholding in the ROIs shown in Figure 4.6a
Average mean transit time (s)

Tissue type Equation (2) Equation (5)
Lateral normal cortical gray matter (ROI 1–ROI 2) 3.93 2.87
Lateral normal subcortical white matter (ROI 1–ROI 2) 4.04 2.71
Medial normal cortical gray matter (ROI 2) 3.96 2.95
Medial normal subcortical white matter (ROI 2) 3.82 2.66
Enhancing primary tumor (ROI 3) 4.38 3.24
Non-enhancing primary tumor (ROI 3) 4.02 2.91
Enhancing metastasis (ROI 4) 4.36 2.99
Non-enhancing metastasis (ROI 4) 4.05 2.81
Enhancing metastasis (ROI 5) 5.02 3.30
Non-enhancing metastasis (ROI 5) 3.30 2.51
9781842143094-Ch04 8/8/07 11:55 AM Page 72
REFERENCES
1. Fiorella D, Heiserman J, Prenger E, Partovi S. Assessment of the reproducibility

of postprocessing dynamic CT perfusion data. Am J Neuroradiol 2004; 25: 97–107.
2. Kudo K, Terae S, Katoh C et al. Quantitative cerebral blood flow measurement
with dynamic perfusion CT using the vascular-pixel elimination method: com-
parison with H2(15)O positron emission tomography. AJNR Am J Neuroradiol
2003; 24: 419–26.
3. Sase S, Honda M, Machida K, Seiki Y. Comparison of cerebral blood flow
between perfusion computed tomography and xenon-enhanced computed
tomography for normal subjects: territorial analysis. J Comput Assist Tomogr
2005; 29: 270–7.
4. Axel L. Cerebral blood flow determination by rapid sequence computed tomog-
raphy. Radiology 1980; 137: 679–86.
9781842143094-Ch05 8/8/07 11:56 AM Page 73
5
Angiogenesis, tumor perfusion,
and cancer management
William W Li, Aliya Jiwani, and Michelle Hutnik
INTRODUCTION
The tumor vasculature is an important target for both cancer therapy

and imaging. Angiogenesis, the growth of new capillary blood vessels,
is the process by which cancer cells recruit their vasculature. Tumor
growth is dependent upon angiogenesis, because new vessels provide
circulatory perfusion and deliver a host of survival factors to proliferat-
ing cancer cells. The angiogenic process in cancer shares similarities
with neovascularization occurring under physiological situations, e.g.
wound healing (granulation), or female reproductive function
(endometrial regeneration, corpus luteum formation, placentation),
but possesses unique features that distinguish the tumor vasculature
from normal blood vessels. The tumor vasculature is characterized
by aberrant vascular architecture, hyperpermeability, disruptions in
arterial–venous hierarchy, and regional areas of stasis. Excessive
vascular permeability leads to intratumoral edema and increased intra-
tumor interstitial pressure, resulting in forces that compress the tumor
vasculature and limit perfusion, as well as the intravascular delivery of
cancer drugs.
Antiangiogenic therapy is a strategy for cancer management first
proposed by Folkman in 1971.1 By targeting the vascular endothelium
supporting tumor growth, angiogenesis inhibition can suppress incipi-
ent tumors, prevent expansion of established tumors, and limit metasta-
tic disease. Some angiogenesis inhibitors have been shown to regress
large tumors to a dormant, microscopic state.2 Antiangiogenic therapy
for cancer became formally established in 2004, with the successful
clinical development and regulatory approval of bevacizumab
(Avastin®), a therapeutic monoclonal antibody designed to target
9781842143094-Ch05 8/8/07 11:56 AM Page 74
vascular endothelial growth factor (VEGF), a potent angiogenic protein

generated by many cancers. Subsequently, other pioneering antiangio-
genic agents approved for oncology include: lenalidomide (Revlimid®),
recombinant human endostatin (EndostarTM), sorafenib (Nexavar®),
sunitinib (Sutent®), and thalidomide (Thalomid®). All have shown sig-
nificant clinical benefit in large-scale clinical trials when administered
for specific cancer indications, either as monotherapy or in combination
with conventional therapy.
Because antiangiogenic agents are in clinical use and development,
imaging the tumor vasculature plays an important role in modern
oncology. These agents alter tumor perfusion and are not always
directly cytotoxic to cancer cells, so imaging tumor mass alone is nei-
ther a reliable nor a sensitive measure of their pharmacological effect.
Instead, imaging modalities that capture features of the tumor vascula-
ture are being highly sought, as they can provide more sensitive and
accurate indicators of pharmacological activity.
This chapter is intended to provide radiologists and imaging special-
ists with an overview of: (1) the basic scientific principles of tumor
angiogenesis; (2) current antiangiogenic strategies for cancer manage-
ment; and (3) new paradigms that are emerging for imaging the tumor
vasculature.
ANGIOGENESIS IN TUMOR GROWTH, PROGRESSION,

AND METASTASES
The incipient stage of all solid tumors is growth restriction due to their
intrinsic lack of a supportive blood supply.1,3,4 At the 60–80-cell stage,
such tumors often grow marginally towards nearby blood vessels
through a process termed vessel cooption.5 Some malignant cells may
infiltrate these local blood vessels to form a ‘mosaic’ vessel comprising
normal vascular cells interspersed with a few tumor cells.6 Vascular
cooption serves the tumor periphery, so tumor expansion leads to
increasing central hypoxia, limiting tumor size to approximately
2–5 millimeters in diameter, until neovascularization takes place. This
initial ‘prevascular’ phase of limited tumor growth may persist for years
without clinical signs or detection.7,8
The rapid phase of tumor growth occurs following the initiation of
tumor angiogenesis, a phenomenon known as the ‘angiogenic switch’,
in which a subset of cancer cells begins secreting angiogenic stimula-
tory factors in excess of inhibitors, thus activating local endothelial cells
to proliferate.9,10 Both human and experimental cancers produce
9781842143094-Ch05 8/8/07 11:56 AM Page 75
Angiogenesis, tumor perfusion, and cancer management 75
numerous peptide proangiogenic factors, such as VEGF, platelet-

derived growth factor (PDGF), fibroblast growth factors (FGFs), and
many other stimulatory mediators.11–13 In human cancer patients, such
factors can be detected in situ within tumor specimens, circulating in
the serum and plasma and cerebrospinal fluid, and excreted in urine.
Other influences in the tumor microenvironment, such as hypoxia,
acidosis, and inflammation, may amplify angiogenic factor expression.
It is now known that the establishment of the angiogenic phenotype
requires tumors to produce angiogenic factors in excess of local
concentrations of endogenous angiogenic inhibitors. Numerous such
inhibitors have been identified, including thrombospondin-1, angio-
statin, endostatin, pigment epithelium-derived factor, vasostatin,
canstatin, and others.8,14 While the mechanisms of action of many
endogenous inhibitors remain poorly defined, they appear to maintain
the quiescent state of endothelial cells under physiological conditions.
Genetic changes in the tumor cell or response to stresses in the
microenvironment lead to the sustained generation of tumor-derived
proangiogenic mediators. A shift in the balance between antiangiogenic
and proangiogenic mediators results, leading to tumor vascularization.
Tumor cells secrete angiogenic factors that promote vascular growth,
while endothelial cells supply oxygen and survival factors that drive
tumor cell growth. Experimental tumor studies have demonstrated that
the onset of angiogenesis may facilitate an increase of tumor volume of
up to 16 000 times in just 2 weeks.
The specific molecular and cellular steps of angiogenesis have been
elucidated: (1) angiogenic factors bind to receptors on endothelial cells
in nearby venules, leading to activation of signal transduction pathways;
(2) endothelial cells secrete proteases that degrade the perivascular
basement membrane; (3) endothelial proliferation occurs; (4) endothe-
lial cells begin to migrate directionally towards the stimulus, facilitated
by further protease secretion at the tips of sprouting vessels and by adhe-
sion molecules that attach endothelial cells to the surrounding extra-
cellular matrix; (5) vascular loops form; (6) arteriovenous differentiation
occurs; and (7) recruitment of mesenchymal cells (pericytes, smooth
muscle cells) to the growing vessel wall aids in stabilization of the vas-
cular structure.15–18 Recently, the contribution of bone marrow-derived
endothelial progenitor cells (EPCs) has been described in tumor angio-
genesis.19–21 Certain angiogenic factors, such as VEGF and placental
growth factor (PlGF), mobilize these cells from bone marrow stem
reservoirs into the circulation. The EPCs migrate to sites of tumor angio-
genesis and incorporate themselves as a small but relevant fraction of
the total endothelium constituting the angiogenic vessels.
9781842143094-Ch05 8/8/07 11:56 AM Page 76
CHARACTERISTICS OF TUMOR ANGIOGENESIS

AND PERFUSION
Tumor vasculature and perfusion are strikingly different from normal

vessels and blood flow. Many of these distinguishing features can be
examined by existing imaging modalities. Malignant vessels are struc-
turally and spatially highly chaotic, with abnormal branching character-
istics22,23 (Figure 5.1). They are disorganized and tortuous, with minimal
vascular hierarchy, and they possess many arteriovenous shunts.24 The
tumor vasculature is hyperpermeable. It can contain pores as large as
2 µm, and is generally an order of magnitude more permeable than
normal vessels, resulting in high intratumoral interstitial pressure and
tumor edema.25
Tumor vessel hyperpermeability is due to significant abnormalities in
the vessel wall ultrastructure.24,25 In an MCa-IV mouse mammary carci-
noma model, openings between (1.7 µm average diameter) and holes
through (0.6 µm average diameter) vessel lining cells were observed.26
Furthermore, in this model, tumor vessel lining cells were irregularly
shaped, and some had cytoplasmic projections measuring up to 50 µm
in length (versus < 1 µm normally observed).26 Malignant vessels gener-
ally have abnormal basement membranes and pericyte coats that are only
loosely associated with endothelial cells and can have multiple layers.27
The vascular abnormalities in tumors underlie the significant hetero-
geneity evident in tumor perfusion. Blood flow can vary considerably
among different regions of the same tumor.28 In BA 1112 sarcomas of
WAG inbred Rijswijk rats, the perfusion rate constant was 15–18 ml/
min/100 g in the inward side of the tumor and only 2–4 ml/min/100 g
a b
Figure 5.1 Scanning electron micrographs of corrosion casts of normal colon

capillaries (a) vs. colon tumor vasculature (b). (Courtesy of Univ.-Prof. Dr med.
Moritz A Konerding of the Institute of Anatomy and Cell Biology, Johannes
Gutenberg-University Mainz, Germany)
9781842143094-Ch05 8/8/07 11:56 AM Page 77
in the center of the tumor.29 In general, blood flow in tumor vessels can
be variable and sluggish.29 It can temporarily stop or even reverse direc-
tion in sections of the tumor.27 Consequently, the tumor vasculature is
inefficient in the delivery of nutrients and oxygen to hypoxic regions,
leading to hypoxia-induced generation of angiogenic factors (i.e. VEGF)
and further rounds of tumor angiogenesis.27
CANCER MANAGEMENT WITH ANTIANGIOGENIC THERAPY
Clinical application of inhibitors of angiogenesis has demonstrated

clinical benefit in cancer patients. A monoclonal antibody, bevacizumab,
directed against VEGF-A, improves overall survival and tumor response,
when administered with conventional 5-fluorouracil (5FU)-based
chemotherapy to patients with advanced colorectal cancer.30 Pre- and
post-treatment tumor biopsies clearly demonstrate a reduction of
microvessel density with bevacizumab treatment.31 Because VEGF also
mediates vascular permeability, anti-VEGF therapy also decreases vessel
leakiness and therefore intratumoral pressure. Clinical studies in
patients with locally advanced rectal cancer have shown a substantial
reduction in tumor interstitial pressure following bevazicumab.31 This
decrease in intramural pressure can allow increased penetration of
chemotherapy agents to the tumor. In theory, agents targeting VEGF
improve the delivery of chemotherapy drugs by pruning away imma-
ture, poorly functional, leaky vessels, thereby ‘unpacking’ remaining
functional vessels by diminishing tumor edema.32 The administration
of bevacizumab in conjunction with chemotherapy has also shown
evidence of clinical benefit in the treatment of non-small cell lung
cancer, metastatic breast cancer, advanced renal cell carcinoma, and
recurrent ovarian cancer.33–35
Small molecule inhibitors of signal transduction have also been
approved for use as antiangiogenic therapy of cancer. Sorafenib is an
oral multikinase inhibitor that improves progression-free survival when
given as monotherapy in patients with advanced, refractory renal cell
carcinoma (RCC). This agent inhibits tyrosine kinase phosphorylation of
the receptors for VEGF (-R1, -R2, -R3), PDGF-Rβ, KIT, Flt-3, and raf
kinase. Because these receptors are present on both malignant cells and
vascular endothelial cells, sorafenib’s effects are directly towards both
cellular compartments within tumors. In recent clinical trials of sorafenib
in advanced RCC, while fewer than 5% of patients had a partial response
to sorafenib as measured with RECIST (Response Evaluation Criteria in
Solid Tumors) criteria, time to progression doubled, suggesting that
9781842143094-Ch05 8/8/07 11:56 AM Page 78
clinical benefit by this agent is derived primarily through disease

stabilization.36–38 This agent is under investigation in melanoma, and
hepatocellular, prostate, thyroid, and non-small-cell lung cancer.
Sunitinib is another oral multikinase inhibitor that targets VEGF-R,
PDGF-Rβ, KIT, Flt-3, and RET. Phase III clinical trials in advanced renal
cell carcinoma demonstrate that sunitinib significantly improves pro-
gression-free survival and response rate (31% in treated group vs. 6% in
placebo group).39 Sunitinib also improves overall survival in patients
with gastrointestinal stromal tumors (GIST) that are refractory to
imatinib (Gleevec®).40 Studies are underway with sunitinib in breast
cancer, non-small cell lung cancer, and prostate cancer.
Antiangiogenic agents have also been approved for hematological
malignancies. Angiogenesis occurs in the bone marrow in such condi-
tions, and proliferating endothelial cells are thought to provide crucial
survival factors to malignant cells.41 Thalidomide with dexamethasone
is approved for use in patients with newly diagnosed multiple myeloma,
in which the combination demonstrates superior response rates com-
pared to dexamethasone alone.42 Lenalidomide is an oral analog of
thalidomide, with potent anticytokine and antiangiogenic properties.
It has been approved for the treatment of low- or intermediate-risk
myelodysplastic syndrome (MDS) associated with the deletion 5q cyto-
genetic abnormality.43 A second approved indication for this agent is
multiple myeloma, where lenalidomide is administered with dexa-
methasone as a second line treatment.44
IMAGING ANGIOGENESIS
One of the major challenges of developing the first generation of

antiangiogenic agents has been the determination of validated pharma-
codynamic markers for efficacy. Angiogenesis imaging is key in evalu-
ating efficacy, from early preclinical drug development through to the
clinical monitoring of patients.18 Cancer patients often have a large
tumor burden with an extensive vascular supply, and imaging is critical
for measuring disease progression. Angiogenesis imaging modalities, in
order to be capable of capturing parameters reflective of tumor vascu-
larity, must be able to accurately quantify small changes against a
potentially large signal background in a sensitive and specific manner.18
In addition to disease progression, tumor vascular imaging has poten-
tial for cancer screening and the detection of early recurrence.45
Parameters associated with tumor angiogenesis include tumor blood
flow, blood volume, permeability, vascular density, and metabolism.18
9781842143094-Ch05 8/8/07 11:56 AM Page 79
The small size (< 100 µm) of tumor microvessels precludes direct
visualization with conventional angiography.26 Current clinical imaging
modalities provide much higher-resolution images than microscopic
methods originally used to study tumor vasculature.27 The combination
of high-resolution imaging with the ability to monitor dynamic (func-
tional) processes has produced major advances in observing and
measuring tumor vasculature. The optimization, validation, and stan-
dardization of these angiogenesis imaging procedures are the next steps
required to accurately compare and consistently reproduce data.46
Coordinated efforts between radiologists, nuclear medicine specialists,
surgeons, medical oncologists, radiation oncologists, and pathologists
will be paramount to the understanding, acceptance, and application of
new imaging techniques.47
Imaging modalities that have been used and are in development to
study vascular features in tumor include high-resolution angiography,
functional computer tomography (CT) imaging, dynamic contrast-
enhanced magnetic resonance imaging (DCE-MRI), high-resolution
magnetic resonance angiography (MRA), positron emission tomography
(PET), Doppler ultrasound (US), and contrast-enhanced US.48,49
Functional CT
CT imaging can be performed with contrast agents to define the
intravascular compartment, including blood flow, blood volume, tissue
perfusion, capillary permeability, and leakage.48,50 Functional CT tech-
niques can delineate increases in tissue perfusion that may reflect
malignancy, even when there is no gross anatomical abnormality
present.50–52
DCE-MRI
MRI can define both blood volume and blood vessel permeability by
using dynamic enhancement of blood pool contrast agents.53,54 The
use of gadolinium can assist in distinguishing between normal (physi-
ological permeability) and malignant (hyperpermeable) tissues,
although the rapid exchange of water between plasma and cell
membranes can lead to an overestimation of the volume of the tissue
compartment by contrast enhancement. Contrast material uptake also
correlates with microvessel density in experimental tumors.
Administration of an anti-VEGF monoclonal antibody to experimental
breast cancers in mice causes decreased vascular permeability that is
detectable with MR imaging.27,55
9781842143094-Ch05 8/8/07 11:56 AM Page 80
MRA
Whereas DCE-MRI focuses on the tumor microvasculature, high-resolution
MRA can be used to view the macroscopic tumor vascular network,
specifically the morphology of larger tumor vessels in the brain.56–59
Initial studies of abnormalities of the larger vessel shape, or tortuosity,
have found a correlation between tumor vessel tortuosity and whether
the tumor is benign or malignant.58
PET
PET imaging is used to evaluate tumor metabolism, as well as blood
flow and volume.60,61 A number of radiotracers, such as water labelled
with oxygen-15, carbon monoxide with carbon-11, and fluorodeoxyglu-
cose with fluorine-18 (18FDG), are available to characterize neoplastic
tissue. By interfering with tumor blood flow, antiangiogenic agents
decrease tumor metabolism. Studies using radiolabeled fluoromisonida-
zole together with PET have been carried out to quantify hypoxia in the
rat glioma and provide functional information about the results of
antiangiogenic therapy.62 A number of novel tracers targeting vascular
integrins have shown experimental utility in characterizing tumor
angiogenesis using PET imaging.63–65
Ultrasound
Ultrasound (US) imaging can identify vascular features in tumors at dif-
ferent levels of resolution (40–200-µm diameter vessels), depending on
the technique employed.66 Contrast agents have dramatically improved
the spatial resolution of US.48 The microbubbles used as contrast agents
for US are generally intravascular and not interstitial, and therefore an
index of blood flow, blood volume, or vascularity within malignant
tissue can be generated. Color flow Doppler US, with or without con-
trast agents, has been used to characterize tumor xenografts in mice
and solid tumors in patients.67–69
IMAGING TUMOR VASCULATURE FOR ANTIANGIOGENIC

AGENT DEVELOPMENT
Imaging angiogenic parameters can provide important early indicators

of antiangiogenic drug efficacy in preclinical studies and in early
9781842143094-Ch05 8/8/07 11:56 AM Page 81
clinical trials. This information can be used to modify the trajectory of

the drug’s development and its treatment approach. As an example,
imaging has been an integral part of the development of sunitinib (also
known as SU11248, Sutent).70–72 DCE-MRI and PET imaging have been
employed to characterize the early pharmacodynamic activity of suni-
tinib in experimental tumors and their vasculature.70 DCE-MRI detected
a 42% decrease in the vascular permeability at the tumor rim in
SU11248-treated experimental mice bearing human colon carcinoma.
Serial PET imaging was also used to detect early clinical activity of
SU11248 in a pilot study for patients with advanced malignancies.72
In that study, H215O-PET imaging demonstrated decreased tumor perfu-
sion in metastatic lesions in patients while treated with SU11248.72
A decrease in the uptake of radiotracer 18FDG was observed, indicating
reduced tumor metabolic rate in treated patients. In patients with gas-
trointestinal stromal tumors (GIST), PET scans of 18FDG avidity dis-
played a rebound pattern.71 There was a suppression of tumor related
18
FDG avidity during treatment with SU11248, followed by a rebound
in 18FDG tumor uptake off treatment. In a long-term evaluation
(> 2 years) of four patients undergoing SU11248 treatment, two patients
continued to show this rebound effect, while two patients no longer
had 18FDG tumor uptake, even off treatment, indicating possible
different subtypes of GIST in these patients.
The successful use of PET and CT scans in evaluating the efficacy of
other approved antiangiogenic agents has been similarly investigated
in various solid tumors, including bevacizumab and sorafenib.31,73,74
Results of PET scans of 18FDG for thalidomide in a preclinical lung
carcinoma study were less conclusive.75 For angiogenesis agents in
development, examples where efficacy was evaluated using DCE-MRI
include VEGFR-2 tyrosine kinase inhibitor SU5416 in a phase I study in
patients with advanced solid tumors,76 and VEGFR tyrosine inhibitor
vatalanib (PTK787/ZK 222584) in patients with advanced colorectal
cancer and liver metastases.77
Because many antiangiogenic agents demonstrate efficacy at levels
well below the maximum tolerated dose (MTD), imaging methodolo-
gies may be applied to determine dose–effect relationships at the
vascular level. Imaging can help to identify the biologically active dose
by quantifying biological effects (i.e. tumor perfusion) and creating cor-
responding dose–response curves.78 An example of this application was
first shown using DCE-MRI in the clinical studies of vatalanib.77,79
Importantly, selection of the optimal biological dose for antiangiogenic
agents may increase efficacy and decrease toxicities.
9781842143094-Ch05 8/8/07 11:56 AM Page 82
CONCLUSIONS
As antiangiogenic therapy for cancer comes of age, the need to

establish validated imaging methodologies for assessing the tumor vas-
culature and its pharmacological alterations has become paramount.
Magnetic resonance and PET imaging are the most common modalities
used for angiogenesis imaging in clinical investigations. Quantitative
methods to evaluate changes in tumor vascular parameters will allow
clinicians to follow the effects of antiangiogenic agents, and to identify
individual patients who are responders to specific treatments versus
non-responders. As the number of angiogenesis inhibitors available
in oncology practice increases, such applications will enable oncolo-
gists to select or change the agent(s) that benefit individual patients.
Patients who are scheduled for serial follow-up imaging during therapy
or while in remission can be assessed for vascular changes indicating
progression or recurrence. Present-day imaging technologies are
capable of being adapted for such purposes, although well-designed
clinical studies must be conducted to establish and validate specific
techniques.
Angiogenesis imaging can clearly aid the drug development process,
enabling it to become more efficient. Preclinical imaging systems will
enable the evaluation of drug candidates, and improve the selection of
lead compounds, at the drug discovery stage. The inclusion of angio-
genesis imaging methodologies in clinical protocols can help to estab-
lish pharmacodynamic activity and dose optimization in early-stage
clinical trials. As new therapeutic targets, such as vascular integrins, are
identified, specific tracers can be designed to selectively target the same
moiety for imaging purposes. Alternatively, the downstream effects of
angiogenesis inhibition in tumors can be assessed, for example, by
imaging annexin V, a marker of endothelial cell apoptosis. Because the
tumor vasculature is heterogeneous and lacks architectural stability, the
selective imaging of angiopoietin-1 (required for vascular stabilization)
or angiopoietin-2 (required for destabilization), with the use of cova-
lently linked monoclonal antibodies to paramagnetic particles, may be
employed to localize and evaluate the state of the tumor vasculature.
Such applications of angiogenesis imaging will require a coordinated
effort between drug and imaging researchers.
Finally, antiangiogenic agents may eventually play a significant role
in cancer chemoprevention, or long-term maintenance therapy during
clinical remission. In this setting, non-invasive imaging of angiogenesis
may be useful as a screening system for incipient cancers, or for
monitoring patients for early signs of recurrence. Clearly, the future of
9781842143094-Ch05 8/8/07 11:56 AM Page 83
radiological investigation and technology development for imaging

angiogenesis is compelling and bright.
ACKNOWLEDGMENTS
The authors would like to acknowledge Vincent W Li, MD and Colleen

Carney of the Angiogenesis Foundation; and Univ.-Prof. Dr med. Moritz
A Konerding of the Institute of Anatomy and Cell Biology, Johannes
Gutenberg-University Mainz, Germany.
REFERENCES
1. Folkman J. Tumor angiogenesis: therapeutic implications. N Engl J Med 1971;

285: 1182–6.
2. Boehm T, Folkman J, Browder T et al. Antiangiogenic therapy of experimen-
tal cancer does not induce acquired drug resistance. Nature 1997; 390: 404–7.
3. Folkman J. Incipient angiogenesis. J Natl Cancer Inst 2000; 92: 94–5.
4. Li CY, Shan S, Huang Q et al. Initial stages of tumor cell-induced angiogene-
sis: evaluation via skin window chambers in rodent models. J Natl Cancer Inst
2000; 92: 143–7.
5. Holash J, Maisonpierre PC, Compton D et al. Vessel cooption, regression, and
growth in tumors mediated by angiopoietins and VEGF. Science 1999; 284:
1994–8.
6. Chang YS, di Tomaso E, McDonald DM et al. Mosaic blood vessels in tumors:
frequency of cancer cells in contact with flowing blood. Proc Natl Acad Sci
USA 2000; 97: 14608–13.
7. Black WC, Welch HG. Advances in diagnostic imaging and overestimations of
disease prevalence and the benefits of therapy. N Engl J Med 1993; 328:
1237–43.
8. Folkman J. Endogenous angiogenesis inhibitors. APMIS 2004; 112: 496–507.
9. Hanahan D, Folkman J. Patterns and emerging mechanisms of the angiogenic
switch during tumorigenesis. Cell 1996; 86: 353–64.
10. Im E and Kazlauskas A. Regulating angiogenesis at the level of PtdIns-4,5-P2.
EMBO J 2006; 25: 2075–82.
11. Folkman J. Role of angiogenesis in tumor growth and metastasis. Semin Oncol
2002; 29: 15–18.
12. Folkman J and Klagsbrun M. Angiogenic factors. Science 1987; 235: 442–7.
13. Milkiewicz M, Ispanovic E, Doyle JL et al. Regulators of angiogenesis and
strategies for their therapeutic manipulation. Int J Biochem Cell Biol 2006; 38:
333–57.
14. Sund M, Hamano Y, Sugimoto H et al. Function of endogenous inhibitors of
angiogenesis as endothelium-specific tumor suppressors. Proc Natl Acad Sci
USA 2005; 102: 2934–9.
15. Folkman J. Angiogenesis. Annu Rev Med 2006; 57: 1–18.
16. Folkman J and Haudenschild C. Angiogenesis in vitro. Nature 1980; 288: 551–6.
9781842143094-Ch05 8/8/07 11:56 AM Page 84
17. Folkman J and Haudenschild C. Angiogenesis by capillary endothelial cells in

culture. Trans Ophthalmol Soc UK 1980; 100: 346–53.
18. Li WW. Tumor angiogenesis: molecular pathology, therapeutic targeting, and
imaging. Acad Radiol 2000; 7: 800–11.
19. Arbab AS, Pandit SD, Anderson SA et al. Magnetic resonance imaging and con-
focal microscopy studies of magnetically labeled endothelial progenitor cells
trafficking to sites of tumor angiogenesis. Stem Cells 2006; 24: 671–8.
20. Asahara T, Masuda H, Takahashi T et al. Bone marrow origin of endothelial
progenitor cells responsible for postnatal vasculogenesis in physiological and
pathological neovascularization. Circ Res 1999; 85: 221–8.
21. Li B, Sharpe EE, Maupin AB et al. VEGF and PlGF promote adult vasculo-
genesis by enhancing EPC recruitment and vessel formation at the site of
tumor neovascularization. FASEB J 2006; 20: 1495–7.
22. Dvorak HF. Angiogenesis: update 2005. J Thromb Haemost 2005; 3: 1835–42.
23. Konerding MA, Fait E, Gaumann A. 3D microvascular architecture of pre-
cancerous lesions and invasive carcinomas of the colon. Br J Cancer. 2001; 84:
1354–62.
24. Raghunand N, Gatenby RA and Gillies RJ. Microenvironmental and cellular
consequences of altered blood flow in tumours. Br J Radiol 2003; 76 (Spec No
1): S11–22.
25. McDonald DM and Baluk P. Significance of blood vessel leakiness in cancer.
Cancer Res 2002; 62: 5381–5.
26. Hashizume H, Baluk P, Morikawa S et al. Openings between defective endothe-
lial cells explain tumor vessel leakiness. Am J Pathol 2000; 156: 1363–80.
27. McDonald DM and Choyke PL. Imaging of angiogenesis: from microscope to
clinic. Nat Med 2003; 9: 713–25.
28. Jain RK. Transport of molecules, particles, and cells in solid tumors. Annu Rev
Biomed Eng 1999; 1: 241–63.
29. Endrich B, Reinhold HS, Gross JF et al. Tissue perfusion inhomogeneity during
early tumor growth in rats. J Natl Cancer Inst 1979; 62: 387–95.
30. Hurwitz H, Fehrenbacher L, Novotny W et al. Bevacizumab plus irinotecan,
fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med
2004; 350: 2335–42.
31. Willett CG, Boucher Y, di Tomaso E et al. Direct evidence that the VEGF-
specific antibody bevacizumab has antivascular effects in human rectal cancer.
Nat Med 2004; 10: 145–7.
32. Jain RK. Normalization of tumor vasculature: an emerging concept in anti-
angiogenic therapy. Science 2005; 307: 58–62.
33. de Gramont A and Van Cutsem E. Investigating the potential of bevacizumab
in other indications: metastatic renal cell, non-small cell lung, pancreatic and
breast cancer. Oncology 2005; 69 (Suppl 3): 46–56.
34. Monk BJ, Han E, Josephs-Cowan CA et al. Salvage bevacizumab (rhuMAB
VEGF)-based therapy after multiple prior cytotoxic regimens in advanced
refractory epithelial ovarian cancer. Gynecol Oncol 2006; 102: 140–4.
35. Morgensztern D and Govindan R. Clinical trials of antiangiogenic therapy in
non-small cell lung cancer: focus on bevacizumab and ZD6474. Expert Rev
Anticancer Ther 2006; 6: 545–51.
36. Adnane L, Trail PA, Taylor I et al. Sorafenib (BAY 43–9006, Nexavar(®)), a
dual-action inhibitor that targets RAF/MEK/ERK pathway in tumor cells and
9781842143094-Ch05 8/8/07 11:56 AM Page 85
tyrosine kinases VEGFR/PDGFR in tumor vasculature. Methods Enzymol 2005;

407: 597–612.
37. Eisen T, Bukowski RM, Staehler M et al. Randomized Phase III trial of
sorafenib in advanced renal call carcinoma (RCC): impact on survival. J Clin
Oncol 2006; 24 (18 Suppl): 4524 (abstr).
38. Ratain MJ, Eisen T, Stadler WM et al. Phase II placebo-controlled randomized
discontinuation trial of sorafenib in patients with metastatic renal cell carci-
noma. J Clin Oncol 2006; 24: 2505–12.
39. Motzer RJ, Rini BI, Bukowski RM et al. Sunitinib in patients with metastatic
renal cell carcinoma. JAMA 2006; 295: 2516–24.
40. Motzer RJ, Hoosen S, Bello CL et al. Sunitinib malate for the treatment of solid
tumours: a review of current clinical data. Expert Opin Investig Drugs 2006;
15: 553–61.
41. Lundberg LG, Hellstrom-Lindberg E, Kanter-Lewensohn L et al. Angiogenesis
in relation to clinical stage, apoptosis and prognostic score in myelodysplastic
syndromes. Leuk Res 2006; 30: 247–53.
42. Rajkumar SV, Blood E, Vesole D et al. Phase III clinical trial of thalidomide
plus dexamethasone compared with dexamethasone alone in newly diag-
nosed multiple myeloma: a clinical trial coordinated by the Eastern
Cooperative Oncology Group. J Clin Oncol 2006; 24: 431–6.
43. List A, Kurtin S, Roe DJ et al. Efficacy of lenalidomide in myelodysplastic
syndromes. N Engl J Med 2005; 352: 549–57.
44. Richardson PG, Blood E, Mitsiades CS et al. A randomized phase 2 study of
lenalidomide therapy for patients with relapsed or relapsed and refractory
multiple myeloma. Blood 2006; 108: 3458–64.
45. El-Deiry WS, Sigman CC and Kelloff GJ. Imaging and oncologic drug develop-
ment. J Clin Oncol 2006; 24: 3261–73.
46. Miller JC, Pien HH, Sahani D et al. Imaging angiogenesis: applications and
potential for drug development. J Natl Cancer Inst 2005; 97: 172–87.
47. Barentsz J, Takahashi S, Oyen W et al. Commonly used imaging techniques
for diagnosis and staging. J Clin Oncol 2006; 24: 3234–44.
48. Atri M. New technologies and directed agents for applications of cancer imag-
ing. J Clin Oncol 2006; 24: 3299–308.
49. Gee MS, Saunders HM, Lee JC et al. Doppler ultrasound imaging detects
changes in tumor perfusion during antivascular therapy associated with vascu-
lar anatomic alterations. Cancer Res 2001; 61: 2974–82.
51. Phongkitkarun S, Kobayashi S, Kan Z et al. Quantification of angiogenesis by
functional computed tomography in a Matrigel model in rats. Acad Radiol
2004; 11: 573–82.
52. Wang JH, Min PQ, Wang PJ et al. Dynamic CT evaluation of tumor vascular-
ity in renal cell carcinoma. AJR Am J Roentgenol 2006; 186: 1423–30.
53. Brasch RC, Li KC, Husband JE et al. In vivo monitoring of tumor angiogenesis
with MR imaging. Acad Radiol 2000; 7: 812–23.
54. Sorensen AG. Magnetic resonance as a cancer imaging biomarker. J Clin Oncol
2006; 24: 3274–81.
55. Preda A, Novikov V, Moglich M et al. MRI monitoring of Avastin antiangio-
genesis therapy using B22956/1, a new blood pool contrast agent, in an
9781842143094-Ch05 8/8/07 11:56 AM Page 86
experimental model of human cancer. J Magn Reson Imaging 2004; 20:

865–73.
56. Brubaker LM, Bullitt E, Yin C et al. Magnetic resonance angiography visualiza-
tion of abnormal tumor vasculature in genetically engineered mice. Cancer Res
2005; 65: 8218–23.
57. Bullitt E, Ewend MG, Aylward S et al. Abnormal vessel tortuosity as a marker
of treatment response of malignant gliomas: preliminary report. Technol
Cancer Res Treat 2004; 3: 577–84.
58. Bullitt E, Zeng D, Gerig G et al. Vessel tortuosity and brain tumor malignancy:
a blinded study. Acad Radiol 2005; 12: 1232–40.
59. Parikh AH, Smith JK, Ewend MG et al. Correlation of MR perfusion imaging
and vessel tortuosity parameters in assessment of intracranial neoplasms.
Technol Cancer Res Treat 2004; 3: 585–90.
60. Blankenberg FG, Eckelman WC, Strauss HW et al. Role of radionuclide imag-
ing in trials of antiangiogenic therapy. Acad Radiol 2000; 7: 851–67.
61. Weber WA. Positron emission tomography as an imaging biomarker. J Clin
Oncol 2006; 24: 3282–92.
62. Cher LM, Murone C, Lawrentschuk N et al. Correlation of hypoxic cell fraction
and angiogenesis with glucose metabolic rate in gliomas using 18F-fluoro-
misonidazole, 18F-FDG PET, and immunohistochemical studies. J Nucl Med
2006; 47: 410–18.
63. Beer AJ, Haubner R, Sarbia M et al. Positron emission tomography using
[18F]Galacto-RGD identifies the level of integrin alpha(v)beta3 expression in
man. Clin Cancer Res 2006; 12: 3942–9.
64. Cai W, Zhang X, Wu Y et al. A thiol-reactive 18F-labeling agent, N-[2-(4–18F-
fluorobenzamido)ethyl]maleimide, and synthesis of RGD peptide-based tracer
for PET imaging of {alpha}v{beta}3 integrin expression. J Nucl Med 2006; 47:
1172–80.
65. Chen X, Park R, Khankaldyyan V et al. Longitudinal microPET imaging of brain
tumor growth with F-18-labeled RGD peptide. Mol Imaging Biol 2006; 8: 9–15.
66. Ferrara KW, Merritt CR, Burns PN et al. Evaluation of tumor angiogenesis with
US: imaging, Doppler, and contrast agents. Acad Radiol 2000; 7: 824–39.
67. Chen JC, Chen Y, Lin JH et al. Resveratrol suppresses angiogenesis in gliomas:
evaluation by color Doppler ultrasound. Anticancer Res 2006; 26: 1237–45.
68. Iordanescu I, Becker C, Zetter B et al. Tumor vascularity: evaluation in a
murine model with contrast-enhanced color Doppler US effect of angiogene-
sis inhibitors. Radiology 2002; 222: 460–7.
69. Shigeno K, Igawa M, Shiina H et al. Transrectal colour Doppler ultrasonogra-
phy for quantifying angiogenesis in prostate cancer. BJU Int 2003; 91: 223–6.
70. Marzola P, Degrassi A, Calderan L et al. Early antiangiogenic activity of SU11248
evaluated in vivo by dynamic contrast-enhanced magnetic resonance imaging
in an experimental model of colon carcinoma. Clin Cancer Res 2005; 11: 5827–32.
71. Raut C, Van Den Abbeele A, Ramaiya N et al. Patterns of PET response after
long-term sunitinib therapy in patients with imatinib resistant gastrointestinal
stromal tumors. J Clin Oncol 2006; 24 (18 Suppl): 9547(abstr).
72. Scott AM, Mitchel P and O’Keefe GJ. Tumor perfusion during treatment with
SU11248 (sunitinib malate) in patients with advanced malignancies as assessed
by (oxygen-15)-water. Presented at 4th International Symposium on Targeted
Therapies, 16–18 March, 2006, Amsterdam, The Netherlands.
9781842143094-Ch05 8/8/07 11:56 AM Page 87
73. Bruckner H, Hrehovovich V and Degregorio P. Laboratory based low dose

combination chemotherapy + bevacizumab for recurrent refractory and
unresectable gastric cancer. J Clin Oncol 2006; 24 (18 Suppl): 14133(abstr).
74. Clark JW, Eder JP, Ryan D et al. Safety and pharmacokinetics of the dual action
Raf kinase and vascular endothelial growth factor receptor inhibitor, BAY
43–9006, in patients with advanced, refractory solid tumors. Clin Cancer Res
2005; 11: 5472–80.
75. Kim DH, Choe YS, Jung KH et al. Synthesis and evaluation of 4-[(18)F]fluoro-
thalidomide for the in vivo studies of angiogenesis. Nucl Med Biol 2006; 33:
255–62.
76. Dowlati A, Robertson K, Radivoyevitch T et al. Novel phase I dose de-escala-
tion design trial to determine the biological modulatory dose of the antiangio-
genic agent SU5416. Clin Cancer Res 2005; 11: 7938–44.
77. Morgan B, Thomas AL, Drevs J et al. Dynamic contrast-enhanced magnetic
resonance imaging as a biomarker for the pharmacological response of
PTK787/ZK 222584, an inhibitor of the vascular endothelial growth factor
receptor tyrosine kinases, in patients with advanced colorectal cancer and liver
metastases: results from two phase I studies. J Clin Oncol 2003; 21: 3955–64.
78. Padhani AR and Neeman M. Challenges for imaging angiogenesis. Br J Radiol
2001; 74: 886–90.
79. Rehman S and Jayson GC. Molecular imaging of antiangiogenic agents.
Oncologist 2005; 10: 92–103.
9781842143094-Ch05 8/8/07 11:56 AM Page 88
9781842143094-Ch06 8/8/07 4:24 PM Page 89
6
Tumors of the brain,
head, and neck
Part A: CT perfusion imaging in cerebral

neoplasms
Karim Samji and Kenneth A Miles
INTRODUCTION
Despite the advent of magnetic resonance (MR), conventional computed

tomography (CT) remains a routine technique in the management of
patients with cerebral neoplasms, not only for diagnosis and staging but
also in treatment planning and the evaluation of treated patients for sus-
pected recurrence. The success of these anatomical imaging modalities is
based on the enhancement of neoplastic brain tissue with intravenously
administered contrast material. An intact blood–brain barrier normally
excludes the contrast medium from the brain, as its microvessel architec-
ture allows only for the passage of small molecules through its tight junc-
tions and narrow intracellular gaps. In contrast, tumor microvessels have
incomplete basement membranes that are abnormally leaky to circulat-
ing molecules and contrast agents. The rate of transendothelial diffusion
of contrast material is a reflection of the integrity of the microvessel wall.1
Furthermore, contrast material within an increased density of tumor
vessels can augment this extravascular contrast enhancement.
Although conventional imaging is capable of delineating areas of
increased contrast enhancement, the eye cannot differentiate between
the intravascular component and the extravascular component of contrast
enhancement. The use of CT perfusion to evaluate these components
of contrast enhancement separately has the potential to improve the
9781842143094-Ch06 8/8/07 4:24 PM Page 90
delineation of tumors and to provide physiological information about

the status of the tumor vasculature, which in turns reflects tumor biol-
ogy and aggression.2 Although such information may also be available
from pathological examination of tumor tissue, in the case of cerebral
tumors, biopsy is highly invasive with a risk of hemorrhage and infec-
tion, and is limited by sampling error.
Hence, the role of imaging in cerebral neoplasms has begun to shift
to provide information on tumor physiology as well as anatomy. CT
perfusion is one of several techniques that have been developed to per-
form functional imaging in order to overcome the inherent limitations
of anatomical imaging. By considering contrast agents as physiological
indicators, it is possible to calculate values for tissue perfusion, includ-
ing blood flow, blood volume, and capillary permeability and leakage.3
This may, in turn, provide invaluable information that will aid
with diagnosis, treatment, and follow-up in patients with cerebral
neoplasms.
The assessment of the vascular permeability of cerebral tumors was
one of the first reported applications of CT perfusion (Chapter 1). The
use of CT perfusion to study cerebral tumors is also a natural extension
to the technique’s wider application in the assessment of cerebrovascu-
lar disease.4 The brain is largely stationary during image acquisition,
and therefore movement artifacts are considerably less likely than for
other body regions. However, the image thresholds that are commonly
used to remove cerebral blood vessels when performing CT perfusion
for patients suffering stroke may not be appropriate for tumor imaging,
because these thresholds can also eliminate highly vascularized tumors
such as meningiomas.
The study of cerebral tumors with CT perfusion is essentially anal-
ogous to the application of MR in this clinical context. Contrast-
enhanced dynamic MR imaging techniques have proven their value in
the detection, grading, and treatment monitoring of cerebral neo-
plasms.1,5–8 However, CT perfusion has some advantages over MR,
including lower cost and greater availability. Quantification is also sim-
pler than with MR, as the relationship between signal and contrast
agent concentration for CT is linear, and quantitative analysis software
is available commercially.9 CT perfusion measurements also demon-
strate greater reproducibility.10 By using existing technologies and contrast
agents, CT perfusion can be readily incorporated into routine conven-
tional CT examinations, and has the potential to be combined with
positron emission tomography (PET) when performed using integrated
PET–CT systems.
9781842143094-Ch06 8/8/07 4:24 PM Page 91
Tumors of the brain, head, and neck 91
MENINGIOMA
Meningiomas are the most common non-glial primary brain tumors,

accounting for approximately 14–20% of all brain tumors in adults.11,12
Meningiomas have characteristic findings on conventional CT imaging.
Hence, their differentiation from intra-axial tumors is relatively easy. The
typical meningioma is a homogeneous, hemispheric, markedly enhancing
extra-axial mass located over the cerebral convexity, in the parasagittal
region, or arising from the sphenoid wing.13 However, atypical features
such as large meningeal cysts, circular contrast enhancement, intratu-
moral hemorrhage, marked peritumoral edema, and various metaplastic
changes (including fatty transformation) are encountered in up to 15%
of patients and can be particularly misleading.13,14
As compared to normal cerebral tissues, meningiomas show
markedly increased values of blood flow, blood volume, and vascular
permeability on CT perfusion15–17 (Figure 6.1). The first reported study
investigating the use of functional CT in the diagnosis of meningiomas
was performed by Jinkins and Nuri Sener.15 They imaged a total of
11 patients with histologically proven benign meningiomas with a
contrast-enhanced dynamic CT protocol, and constructed dynamic per-
fusion curves based on the relative changes in CT number versus time.
Every patient imaged demonstrated intratumoral neovascularity during
the stage of maximal arterial perfusion, reflective of early extravasation
of contrast medium into the tumor.15 The brain tissue abutting upon the
tumor also demonstrated varying perfusion patterns. The cortex overlying
white matter edema revealed relative hyperperfusion, a phenomenon
that may reflect autoregulation loss and secondary cerebral ‘luxury’
perfusion.15 Hypoperfusion of the cerebrum interposed between the
tumor and a rigid structure such as the calvarium was shown on the
dynamic perfusion curves, compatible with the theory of compressive
cerebral hypoperfusion.15
In a CT perfusion study of six patients with intracranial meningioma
undergoing brachytherapy, Bondestam et al. reported perfusion values
of 231 ± 58 and 224 ± 54 ml/min/100 g in the tumor center and periph-
ery respectively.16 Cheong et al. reported CT perfusion values from
tumor and normal brain in a series of six patients with intracranial
meningioma.17 These researchers compared values for perfusion, frac-
tional volume of tracer distribution within the intravascular and
extravascular spaces, and the permeability surface area (PS) product
obtained by three different kinetic models. Absolute values for these
parameters depended on the model used. Although the distributed
9781842143094-Ch06 8/8/07 4:24 PM Page 92
a b 500 ml/min/100 ml
0 ml/min/ml
c d 50 ml/min/100 ml
20 ml/100 ml
0 ml/100 ml 0 ml/min/100 ml
Figure 6.1 Contrast-enhanced computed tomography (CT) (a), CT perfusion

(b), blood volume (c), and vascular permeability image (d) from a patient with
a right frontal meningioma. Tumor values for perfusion (94 ml/min/100 ml),
relative blood volume (7.2 ml/100 ml), and permeability (18.9 ml/min/100 ml)
are considerably greater than in normal cerebral tissues
parameter model was considered more reliable, there were significant

linear correlations between estimates obtained from all kinetic models,
and normal values from compartmental analysis more closely approxi-
mated equivalent values reported using other techniques.17 Normal
cerebral tissues yielded PS values close to zero, whereas the values in
meningioma were substantially higher due to a disrupted blood–brain
barrier (Table 6.1). Tissue perfusion and the fractional intravascular
volume of tracer (i.e. relative blood volume) were also increased in
meningioma in comparison to normal brain tissue. Hence, the authors
concluded that enhancement of meningioma on dynamic contrast-
enhanced CT could be based on a combination of increased extravasa-
tion of tracer, high perfusion, and dense vasculature.17
9781842143094-Ch06 8/8/07 4:24 PM Page 93
Table 6.1 Computed tomography (CT) perfusion values for normal cerebral tissues and
meningioma as reported by Cheong et al. using compartmental analysis17
Perfusion Relative blood Permeability

(ml/min/100 g) volume (ml/100 g) (ml/min/100 g)
Normal gray matter 58.4 ± 12.5 1.74 ± 0.64 0.17 ± 0.15

Normal white matter 38.8 ± 15.5 1.09 ± 0.29 0.08 ± 0.12
Meningioma 137 ± 33.5 10.7 ± 7.12 52.4 ± 59.3
GLIOMA
Gliomas are astrocytic, oligodendroglial, and ependemal in origin, and

account for 70% of all brain tumors.18 The most common (65%) and
most malignant histological subtype is the glioblastoma, and the 5-year
survival rate for patients with glioblastoma is less than 3%.18 The intro-
duction of CT and MR imaging in recent decades has significantly
improved the care of patients with intracranial gliomas. Novel imaging
techniques can add to the radiologist’s diagnostic armamentarium, and
allow more precise anatomical localization of the tumor as well as pro-
vide a means for assessing tumor grade and measuring its progression
and response to treatment.
Perfusion imaging of brain tumors has been shown to provide valu-
able information about cerebral gliomas.1,2,5,6,8,19–21 Specifically, dynamic
imaging-generated maps of cerebral blood volume (CBV) demonstrated
a statistically significant increase in CBV in glioma when compared to
normal brain tissue.5,8 In addition, disruptions in the integrity of the
blood–brain barrier are noted, as evidenced by an increased PS in glioma
when compared to normal brain tissue.2,21,22 Nabavi et al. and Eastwood
and Provenzale have also documented their finding of increased cerebral
blood flow (CBF) in cerebral gliomas in published case reports.19,20
Although cerebral gliomas exhibit increased perfusion and blood
volume, the values are often only slightly higher than in normal cerebral
tissues2,19,20 (Figures 6.2 and 6.3). For example, the cases reported by
Nabavi et al. and Eastwood and Provenzale yield relative blood volumes
of 5.5 and 2.3 ml/100 g respectively.19,20 On the other hand, these tumors
are readily seen on CT images of vascular permeability which can some-
times reveal areas of tumor not detected by conventional contrast-
enhanced CT (Figure 6.3).
The grading of gliomas is of significant clinical importance, as thera-
peutic options for these tumors differ considerably depending on grade.
9781842143094-Ch06 8/8/07 4:24 PM Page 94
a b
Figure 6.2 Conventional contrast-enhanced CT (a) and CT perfusion (b)

images of a patient with a cerebral glioma (arrow) obtained during stereotactic
imaging. Tumor perfusion values are close to those of cerebral cortex
High grade gliomas are usually treated with adjuvant radiation therapy
or chemotherapy after resection, whereas this is not the case for low
grade gliomas.6 Conventional contrast-enhanced CT can provide only a
limited amount of information concerning the grade of glioma. Thus,
the current standard for tumor grading is histopathological assessment,
but this is complicated by the heterogeneous nature of gliomas, and his-
tological samples obtained at biopsy may be subject to sampling error.8
Hence, there is a need for a more accurate and less invasive method of
tumor grading.
In a study of 22 patients, Ding et al. assessed the ability of CT perfusion
to grade cerebral glioma.21 Both CBV and PS were strongly correlated
with tumor grade, and the differences in CBV and PS were statistically
significant between low- and high-grade glioma.21 In addition, receiver
operating characteristic curves revealed better diagnostic performance for
PS in comparison to CBV for determining glioma grade.21 These findings
are analogous to those reported previously using MR.1,5,6,8
CEREBRAL LYMPHOMA
Primary central nervous system lymphoma exhibits a biological behavior

that is different from that of other primary brain tumors, leading to dif-
ferences in treatment regimen. The appearances of cerebral lymphoma
9781842143094-Ch06 8/8/07 4:24 PM Page 95
Figure 6.3 Conventional contrast-enhanced CT (a), BBB-permeability image

(b) and blood volume image (c) from a patient with recurrent left frontal
glioma. Note the second focus of tumor (arrow) seen on the BBB-permeability
image which is not seen on conventional CT. (Note the normally permeable
choroids plexuses are seen as areas of high permeability posteriorly.)
(Reproduced with permission from reference 23)
9781842143094-Ch06 8/8/07 4:24 PM Page 96
on conventional CT can also differ from other tumors, with most cases
showing homogeneous contrast enhancement. Warnke et al. have used
CT perfusion to study seven patients with cerebral lymphoma.24 As in
the case of other tumors, vascular permeability was markedly increased
compared to normal cerebral tissue (2.95 ± 1.06 ml/min/100 g), and
higher than in other cerebral tumors assessed using the same technique.
However, the relative blood volume of tumor was comparable to that of
normal brain (2.7 ± 2.4 ml/100 g), and tumor perfusion values, measured
in this study by xenon CT, were also close to those for normal white
matter (43.2 ± 10.5 ml/min/100 g).
METASTATIC BRAIN DISEASE
CT perfusion is potentially of value in the diagnosis, management, and

post-therapeutic monitoring of metastatic brain disease.25,26 Small metas-
tases are frequently missed with contrast-enhanced CT.27 However, the
discovery of multiple, small, asymptomatic lesions may direct more
patients towards surgical resection and radiation treatment. This
approach has been shown to control intracranial disease, and may
improve survival and quality of life.27,28
CT perfusion findings in patients with cerebral metastases have been
reported in two cases by Roberts et al.25 In common with other cerebral
tumors, the vascular permeability was significantly raised, up to more
than 40 ml/min/100 g in a case of metastatic rectal cancer. However,
values for tumor perfusion and relative blood volume were much more
variable, in one case close to values in normal cerebral tissues. Although
this technique may be of value in therapeutic monitoring, further studies
are needed to determine the diagnostic accuracy of CT perfusion as
compared to conventional CT for this indication, and whether there are
any long-term benefits from adopting such a protocol.
CT PERFUSION TO DIFFERENTIATE BETWEEN BRAIN

TUMOR AND CEREBRAL INFARCTION
It is occasionally difficult to differentiate between brain tumor and

cerebral infarction during the initial presentation. Multiple investigations,
such as magnetic resonance imaging (MRI) and fluorodeoxyglucose
(FDG)-PET, may be adopted in such circumstances to establish a firm
diagnosis. Cases reported by Keith et al. and Lee et al. illustrate the
potential for CT perfusion to provide a readily available and low-cost
method to differentiate between these two pathologies (Figure 6.4).29,30
9781842143094-Ch06 8/8/07 4:24 PM Page 97
CT CBV CBF
Figure 6.4 Conventional CT and perfusion CT images (cerebral blood volume

(CBV) and cerebral blood flow (CBF)) from a 56-year-old man presenting with
a sudden onset of expressive dysphasia followed by focal right-sided motor
seizure activity. The conventional CT demonstrates a hypodense area in the left
hemisphere (arrow) which was initially considered to be an early cerebral
infarct. However, CT perfusion images demonstrate a slight increase in blood
flow in the corresponding area, implying low-grade tumor. (Reproduced with
permission from reference 29)
Cerebral tumors typically exhibit CBF and CBV values that are similar to
or higher than those obtained from normal brain tissue. In contrast, non-
hemorrhagic stroke is readily identified on CT perfusion as an area of
reduced CBF.29 CBV may be normal in the presence of reversible ischemia,
reflecting preservation of vascular autoregulation, whereas irreversible infarc-
tion is characterized by matched reductions in CBF and CBV.3 However,
images of vascular permeability are less likely to differentiate cerebral tumor
and infarction, as values are increased above normal in both pathologies.
THERAPEUTIC PLANNING AND MONITORING
Radiation planning
The management of malignant cerebral neoplasms requires aggressive
treatment modalities including surgery and radiotherapy, but the prognosis
of these tumors still remains poor. In order to maximize the radiation
dose to the tumor and minimize the damage to normal surrounding
tissue, an accurate and reliable identification of viable tissue margins
needs to be obtained.31 Anatomic images may not sufficiently identify
eloquent cortex, and may misjudge the distance from potential treatment
margins. Incorporation of physiological imaging into the planning
process may improve tumor targeting by more accurate target definition
and reduction of the total target volume. Furthermore, there is interest in
9781842143094-Ch06 8/8/07 4:24 PM Page 98
a b
Figure 6.5 Tumor target volumes constructed by a radiation oncologist on the

basis of conventional CT (a) and permeability CT (b) images. The improved
delineation of the tumor on the permeability image has resulted in a change in
target volume
using physiological imaging to identify the most active areas of tumor

which can then be selected for more intense radiotherapy.
Conventional CT images are widely used as the basis for defining
target volumes for radiotherapy treatment of cerebral tumors. Even when
other functional imaging modalities are incorporated into the planning
process, a CT image is usually also required to provide a map of X-ray
attenuation. CT perfusion can be readily appended to the conventional
CT performed in such circumstances, and the ability for CT-derived
images of vascular permeability to improve the delineation of cerebral
tumors (Figure 6.3) highlights the potential for CT perfusion to con-
tribute to the definition of target volumes. Furthermore, the association
between permeability on CT and tumor grade21 suggests the potential
for permeability maps to be used as a basis for intensity modulation.
A pilot study by Reardon et al. has shown that CT-derived images of
vascular permeability can potentially impact on tumor target volumes
(Figure 6.5). Two radiation oncologists constructed target volumes for
six cerebral tumors using conventional CT and blood–brain barrier
permeability (BBBP) CT images. Target volumes constructed on BBBP
images tended to be smaller (change in size –43% to +33%, median
–19%, p = 0.06) and were displaced by 1.6% to 24% (median 11.5%) of
the mean target volume diameter (Reardon, personal communication).
Further prospective studies are required to assess the impact of this
approach on radiotherapy treatment efficacy and toxicity.
9781842143094-Ch06 8/8/07 4:24 PM Page 99
THERAPEUTIC MONITORING
Conventional CT and MR imaging are routinely used in the follow-up

of patients after radiation treatment or chemotherapy for cerebral neo-
plasms. In the light of the limitations of RECIST (Response Evaluation
Criteria in Solid Tumors) and other anatomically based response criteria,
there is an increasing interest in the application of imaging techniques
that assess tumor physiology for the evaluation of tumor response to
therapy (Chapter 12). In the case of radiotherapy for cerebral tumors,
the situation is further complicated by difficulties in differentiating
between radiation necrosis and recurrent neoplastic growth.
CT perfusion has been successfully used to demonstrate effects of
both chemotherapy and radiotherapy upon cerebral tumors. Changes in
tumor vascular permeability have been shown in response to dexa-
methasone32 and the bradykinin analog RMP-7.22 Radiotherapy
response has been studied in small series of patients with meningioma
undergoing brachytherapy.16 From initial values of 231 ml/min/100 ml,
perfusion in the center of the tumor fell significantly at 3 months
(by 41%) and 1 year (by 68%) following the implantation of iodine
seeds, whereas perfusion in the tumor periphery fell little. Millar et al.
related the changes in CT perfusion parameters in normal brain to
symptoms of toxicity following whole-brain irradiation in 14 patients
with cerebral metastases.26 After an initial 19% increase, PS values fell
to baseline values at day 5. Changes in mean transit time and blood
volume correlated with headache and nausea scores, respectively.
Both radiation necrosis and recurrent tumor can present as an
enhancing brain lesion following radiotherapy. Although no systematic
studies of the ability for CT perfusion to distinguish these pathologies
systematic has been performed, the potential role for perfusion imag-
ing in this context is suggested by the MR perfusion study of Sugahara
et al.34 The ratio of CBV in the lesion compared to that in normal brain
tended to be higher in recurrent tumor than in radiation necrosis
(median values: 2.60 vs. 1.37). Initial experience with CT perfusion has
found very similar results with equivalent CBV ratios of 2.54 and 1.17.
A cut-off value of 1.4 gave a sensitivity of 88.5% and specificity 80% for
the diagnosis of recurrence versus necrosis.
CONCLUSION
CT perfusion is emerging as a viable functional imaging modality of

value as an adjunct to conventional CT in the diagnosis, treatment
9781842143094-Ch06 8/8/07 4:24 PM Page 100
planning, and follow-up of cerebral neoplasms. Although constrained

by the need to use ionizing radiation, the potential for adverse reaction to
contrast agent, and limited anatomic coverage, CT perfusion may prove
to have advantages over MR imaging in the assessment of tumor angio-
genesis. The technique is fast and readily available on most modern
spiral CT scanners equipped with the appropriate software, and can be
easily incorporated into conventional CT examinations performed as
part of the clinical management for many patients with cerebral tumors.
Part B : CT perfusion in head and

neck cancer
Robert Hermans
INTRODUCTION
Beginning with the observations of Gray et al.,35 oxygen is known to

be a powerful radiosensitizer: oxygen enhances the formation of free
radicals or draws existing free radicals into chain reactions, producing
new damaging free radicals; another mechanism postulated is that
many irradiation-induced chemical changes are being blocked by the
presence of oxygen. An imbalance between oxygen supply and con-
sumption, largely resulting from the presence of inadequate and hetero-
geneous vascular networks, leads to chronic tumor hypoxia.36 Tumor
hypoxia is not only a chronic phenomenon, related to vascular density,
in cells which are situated far from the vessels,37 but also an acute cyclic
phenomenon as part of a dynamic process in which the vessels period-
ically open and close. In some tumors, a high interstitial pressure may
limit the diffusion of oxygen towards the cells.
In the second half of the 20th century, studies showed the clinically
relevant effect of tumor hypoxia on the response to radiation therapy.36
Also, in head and neck cancer, several studies found a significantly
worse response to irradiation in tumors with a hypoxic subvolume.37–9
Recent laboratory and clinical data have shown that hypoxia is also
associated with a more malignant phenotype, affecting genomic stability,
apoptosis, angiogenesis, and metastasis.40
Direct quantification of tumor oxygenation can be expected to be of
important prognostic value. Tumor oxygenation has been measured
9781842143094-Ch06 8/8/07 4:24 PM Page 101
invasively using oxygen-sensitive needle electrodes in animal tumors

and in certain human tumors. This technique can be used in head and
neck tumors,39 but many primary tumors in this region are deeply
seated, difficult to reach, and close to critical anatomic structures.
There is a need for a non-invasive method to measure tumor oxy-
genation, not only as a predictor of outcome, but also to select patients
for concomitant radiosensitizing therapy to overcome the hypoxia
effect. For example, it has been reported that the oxygenation of head
and neck tumors improves during carbogen breathing41 or hyperbaric
oxygenation.41 Increased oxygenation of tumors treated with carbogen
and nicotinamide has been demonstrated in patients. Promising results
have been obtained in non-randomized clinical studies using this
combination in conjunction with accelerated irradiation.43,44
Perfusion can be defined as the blood flow through a tissue of inter-
est per unit of volume. Tumor perfusion and tumoral oxygen concen-
tration are factors which are usually strongly linked, although tumor
oxygenation also depends on oxygen consumption by the tumor cells.
The oxygen availability or oxygen supply is the amount of oxygen
carried by the blood to a given tissue per unit of time; it is the product
of the perfusion rate and the arterial oxygen concentration.
The cross-sectional images obtained with CT provide detailed mor-
phologic information. Besides anatomical analysis, functional analysis
of the images is also possible.
CT PERFUSION IMAGING IN HEAD AND NECK CANCER
Few groups have investigated the possibilities of dynamic CT to obtain

measurements of blood flow in head and neck cancer. The data reported
by Hermans et al.45–47 were obtained by using the ‘gradient method’.48–50
Perfusion rates in primary head and neck cancer, as measured by
dynamic CT, were shown to be obtainable with good intra- and inter-
observer reproducibility.46 The mean CT-determined perfusion value,
measured in 105 patients, was 88.8 ml/min/100 g (median 83.5; standard
deviation (SD) 46.5).47
Brix et al.51 reported their findings in six patients suffering head and
neck cancer. These authors used an extension of the gradient method,
taking into account also the bidirectional diffusion of the contrast
medium across the capillary wall by using a two-compartmental model;
this should allow a more comprehensive characterization of tissue
microcirculation. A mean regional blood flow of 45 ± 16 ml/min/100 g
was measured.
9781842143094-Ch06 8/8/07 4:24 PM Page 102
Using a deconvolution-based CT perfusion method based on linear

system theory as described by Jaschke et al.52, average values of tissue
blood volume, blood flow, mean transit time, and capillary permeability
surface area product have been reported in head and neck cancer;53
using this method, the average perfusion rate measured in 14 patients
was 126.2 ml/min/100 g (SD 76.5). Several problems exist when such
methods are applied to dynamic CT: the transit time is difficult to obtain
from the tissue time–density curve, iodinated contrast agents do not
remain purely intravascular, and the latter part of the time–density
curve is difficult to analyze due to recirculation of the contrast medium
and other artifacts.49
USE OF CT PERFUSION AS PREDICTOR OF LOCAL

CONTROL AFTER RADIOTHERAPY
The aim of a study reported by Hermans et al.47 was to test the hypoth-
esis that the outcome of patients is affected by the tumor perfusion rate.
Multivariate analysis confirmed the value of CT-determined tumor per-
fusion as an independent predictor of local control in head and neck
cancer, treated by definitive radiotherapy, with or without adjuvant
chemotherapy. Patients with a low perfusion value showed a statistically
significantly higher local failure rate than those with a high perfusion
value (Figure 6.6). Presumably, this is linked to more extensive and/or a
higher degree of hypoxia in low-perfused tumors, as suggested by others.54
The perfusion rate was also found to be independent of tumor
volume. CT-determined primary tumor volume has been shown to be
an important predictor of local control for several head and neck cancer
sites, including glottic,55 supraglottic,56 hypopharyngeal,57 and nasopha-
ryngeal cancer.58,59 As perfusion rate is independent of tumor volume,
it can be considered an additional parameter helping to predict the out-
come of the patient after treatment.
The perfusion rate was found to be independent of the T-classification.47
Local control was significantly different according to the median perfu-
sion rate in the subgroups T3 and T4 (Figures 6.7 and 6.8). Particularly
in the T4-group, the difference in local outcome between those patients
with high and low perfusion values is very pronounced. This opens
perspectives to use CT perfusion as a tool to select patients suffering
advanced head and neck cancer who may benefit from concomitant
treatment during irradiation.
The practical advantages of this method are the low extra burden to
patients, the ease of implementing it on existing CT machines, and
9781842143094-Ch06 8/8/07 4:24 PM Page 103
1
< 83.5 ml/min/100 g
0.9
> 83.5 ml/min/100 g
0.8
0.7
Local control
0.6
0.5
0.4
0.3
0.2
0.1 p < 0.05
0
0 6 12 18 24 30 36 42
Time (months)
Figure 6.6 Local control over time in 105 patients suffering head and neck
cancer. The patients are stratified into two groups according to the median
primary tumor perfusion value, as determined by dynamic CT using the
gradient method. (Reproduced with permission from reference 47)
1
< 83.5 ml/min/100 g
0.9
> 83.5 ml/min/100 g
0.8
0.7
Local control
0.6
0.5
0.4
0.3
0.2
0.1 p < 0.05
0
0 6 12 18 24
Time (months)
Figure 6.7 Local control versus perfusion rate (patients classified as T3 (n = 39))
plotted over time after start of radiotherapy. (Reproduced with permission from
reference 47)
9781842143094-Ch06 8/8/07 4:24 PM Page 104
1
< 83.5 ml/min/100 g
0.9
> 83.5 ml/min/100 g
0.8
0.7
Local control
0.6
0.5
0.4
0.3
0.2
0.1 p < 0.05
0
0 6 12 18 24
Time (months)
Figure 6.8 Local control versus perfusion rate (patients classified as T4 (n = 34))
plotted over time after start of radiotherapy. (Reproduced with permission from
reference 47)
the low cost compared to other non-invasive and invasive methods to

estimate tumor perfusion or oxygenation.
However, CT perfusion also has a number of shortcomings. An impor-
tant limitation is that only one level through the tumor can be examined;
therefore, the obtained results may not be representative for the entire
tumor. Averaging the perfusion rate in a large region of interest (ROI)
may also not provide representative results, as a few hypoxic clono-
genic cells in a small low-perfusion area may determine the long-term
outcome in spite of a high overall perfusion rate. Furthermore, it may
prove difficult to anatomically define hypoxic subvolumes, of interest
in intensity-modulated radiation therapy.60 However, the recent techni-
cal developments in computed tomography, using multidetector tech-
nology, allow simultaneous examination of several tumor levels;61
combined with advances in software analysis tools, parametric quantifi-
cation of perfusion in the entire tumor volume now becomes feasible
(Figure 6.9). Nevertheless, small movements of the soft tissue structures
in the neck, such as caused by arterial pulsations, breathing, or swal-
lowing, may occur, both in the axial plane and in the long axis of the
patient, rendering them difficult to compensate. Such tissue movements
may make a parametric analysis difficult to realize in all patients.
The presented technique is not well suited to examine the known
temporal heterogeneity of tumor perfusion, as several hours are needed
before the baseline tissue density is restored through renal clearance of
9781842143094-Ch06 8/8/07 4:24 PM Page 105
Figure 6.9 Time frame of a dynamic series obtained during the intravenous
injection of a contrast agent bolus, using a four-row multidectector CT machine.
A pixel-by-pixel analysis of the dynamic image series was performed. The perfusion
data are color-coded and mapped on the anatomical images. The perfusion rate
shows a heterogeneous distribution throughout the tumor (calculated using
Body CT Perfusion; Siemens Medical Solutions, Erlangen, Germany)
the contrast medium; temporal fluctuations in tumor perfusion therefore

cannot be estimated.
Until now, reported perfusion values were estimated in the primary
tumor location only. From a functional point of view, obtaining control
of the primary tumor by (chemo)radiotherapy is more important than
controlling the adenopathies. Residual neck disease after irradiation can
be surgically treated, causing some morbidity, but usually less severe
than surgical resection of the primary tumor.62
CONCLUSION
The perfusion rate of head and neck tumors can be determined by

dynamic CT, during a routine CT study, with little supplementary
9781842143094-Ch06 8/8/07 4:24 PM Page 106
burden for the patient. This parameter has significant predictive value
for local outcome after irradiation with curative intent, and therefore
may be considered a useful tool to select patients for concomitant treat-
ment in advanced head and neck cancer.
REFERENCES
1. Roberts HC, Roberts PL, Brasch RC, Dillon WP. Quantitative measurements of
microvascular permeability in human brain tumors achieved using dynamic
contrast-enhanced MR imaging: correlation with histologic grade. AJNR Am J
Neuroradiol 2000; 21: 891–9.
2. Leggett DAC, Miles KA, Kelley BB. Blood-brain barrier and blood volume
imaging of cerebral glioma using functional CT: a pictorial review. Eur J Radiol
1999; 30: 185–90.
3. Miles KA. Brain perfusion: computed tomography applications. Neuroradiology
2004; 46: S194–S200.
4. Miles KA, Eastwood JD, Konig M, eds. Multidetector Computed Tomography
in Cerebrovascular Disease: CT Perfusion Imaging. Abingdon, UK: informa
Healthcare, 2007.
5. Jackson A, Kassner A, Annesley-Williams D et al. Abnormalities in the recircula-
tion phase of contrast agent bolus passage in cerebral gliomas: comparison with
relative blood volume and tumor grade. AJNR Am J Neuroradiol 2002; 23: 7–14.
6. Law M, Yang S, Babb JS et al. Comparison of cerebral blood volume and
vascular permeability from dynamic susceptibility contrast-enhanced perfusion
MR imaging with glioma grade. AJNR Am J Neuroradiol 2004; 25: 746–55.
7. Provenzale JM, Mukundan S., Dewhirst M. The role of blood-brain barrier
permeability in brain tumor imaging and therapeutics. AJR Am J Roentgenol
2005; 185: 763–7.
8. Aronen HJ, Gazit IE, Louis DN et al. Cerebral blood volume maps of gliomas:
comparison with tumor grade and histologic findings. Radiology 1994; 191: 41–51.
9. Miles, KA, Charnsangavej C, Lee FT et al. Application of CT in the investiga-
tion of angiogenesis in oncology. Acad Radiol 2000; 7: 840–50.
10. Miles KA, Wintermark M. Comparison with other techniques. In: Miles KA,
Eastwood JD, Konig M, eds. Multidetector Computed Tomography in
Cerebrovascular Disease: CT Perfusion Imaging. Abingdon, UK: Informa
Healthcare, 2007; 159–70.
11. Rachlin JR. Etiology and biology of meningiomas. In: Al-Mefty O, ed.
Meningiomas. New York: Raven Press, 1991; 27–35.
12. Russel DS, Rubenstein LJ. Pathology of Tumors of the Central Nervous System,
4th edn. London: Edward Arnold, 1995: 66–91.
13. Buetow MP, Buetow PC, Smirniotopoulos JG. Typical, atypical and misleading
features in meningioma. Radiographics 1991; 11: 1087–106.
14. Hakyemez B, Yildirim N, Gokalp G, Erdogan C, Parlak M. The contribution of
diffusion-weighted MR imaging to distinguish typical from atypical meningiomas.
Neuroradiology 2006; 48: 513–20.
15. Jinkins JR, Nuri Sener R. The characteristics of cerebral meningiomas and
surrounding tissues on dynamic CT. Neuroradiology 1991; 33: 499–506.
9781842143094-Ch06 8/8/07 4:24 PM Page 107
16. Bondestam S, Halavaara JT, Jaaskelainen JE et al. Perfusion CT of the brain

in the assessment of flow alterations during brachytherapy of meningioma.
Acta Radiol 1999; 40: 469–73.
17. Cheong LHD, Lim CCT, Koh TS Dynamic contrast-enhanced CT of intracranial
meningioma: comparison of distributed and compartmental tracer kinetic
models – initial results. Radiology 2004; 232: 921–30.
18. Ohgaki H, Kleihues P. Epidemiology and etiology of gliomas. Acta Neuropathol
2005; 109: 93–108.
19. Nabavi DG, Cenic A, Craen RA et al. CT assessment of cerebral perfusion: exper-
imental validation and initial clinical experience. Radiology 1999; 213: 141–9.
20. Eastwood JD, Provenzale JM. Cerebral blood flow, blood volume, and vascular
permeability of cerebral glioma assessed with dynamic CT perfusion imaging.
Neuroradiology 2003; 45: 373–6.
21. Ding B., Ling HW, Chen KM, Jiang H, Zhu YB. Comparison of cerebral blood
volume and permeability in preoperative grading of intracranial glioma using
CT perfusion imaging. Neuroradiology 2006; 48: 773–81.
22. Ford JM, Miles KA, Hayball MP et al. A simplified technique for measurement
of blood-brain barrier permeability using computed tomography: preliminary
results of the effect of RMP-7. In: Faulkner K, Carey B, Crellin A, Harrison RM,
eds. Quantitative Imaging in Oncology. Proceedings of the 19th LH Gray
Conference. London: British Institute of Radiology, 1996: 1–3.
23. Miles KA, Dawson P. CT permeability imaging: blood-brain barrier imaging.
In: Miles K, Blomley M, Dawson P, eds. Functional Computed Tomography.
Oxford: ISIS Medical Media, 1997: 157–65.
24. Warnke PC, Timmer J, Ostertag CB, Kopitzki K. Capillary physiology and
drug delivery in central nervous system lymphomas. Ann Neurol 2005; 57:
136–9.
25. Roberts HC, Roberts TP, Lee TY, Dillon WP. Dynamic, contrast-enhanced CT
of human brain tumors: quantitative assessment of blood volume, blood flow,
and microvascular permeability: report of two cases. AJNR Am J Neuroradiol
2002; 23: 828–32.
26. Millar BM, Purdie TG, Yeung I et al. Assessing perfusion changes during whole
brain irradiation for patients with cerebral metastases. J Neurooncol 2005; 71:
281–6.
27. Young RJ, Sills AK, Brem S. Neuroimaging of metastatic brain disease.
Neurosurgery 2005; 57 (Suppl 4): 10–23.
28. Sheehan J, Niranjan A, Flickinger JC, Kondziolka D, Lunsford LD. The expand-
ing role of neurosurgeons in the management of brain metastases. Surg Neurol
2004; 62: 32–40.
29. Keith CJ, Griffiths M, Peterson B, Anderson RJ, Miles KA. Computed tomogra-
phy perfusion imaging in acute stroke. Australas Radiol 2002; 46: 221–30.
30. Lee R, Cheung RTF, Hung KN. Use of CT perfusion to differentiate between
brain tumor and cerebral infarction. Cerebrovasc Dis 2004; 18: 77–83.
31. Gross MW, Weber WA, Feldmann HJ et al. The value of F-18-fluorodeoxy-
glucose PET for the 3-D radiation treatment planning of malignant gliomas.
Int J Radiat Oncol Biol Phys 1998; 41: 989–95.
32. Yeung WT, Lee TY, Del Maestro RF et al. Effect of steroids on iopamidol
blood-brain barrier transfer constant and plasma volume in brain tumors meas-
ured with X-ray computed tomography. J Neurooncol 1994; 18: 53–60.
9781842143094-Ch06 8/8/07 4:24 PM Page 108
33. Jain R, Elika S, Searpace L, Schultz L, Mikkelsen T, Pakel S. Perfusion CT: Initial
experience in differentiating recurrent brain tumors from radiation effects/
radiation necrosis. European Radiology 2007; 17: 291.
34. Sugahara T, Korogi Y, Tomiguchi S et al. Posttherapeutic intraaxial brain
tumor: the value of perfusion-sensitive contrast-enhanced MR imaging for
differentiating tumor recurrence from non-neoplastic contrast-enhancing
tissue. Am J Neuroradiol 2000; 21: 901–9.
35. Gray LH, Conger AD, Ebert M et al. The concentration of oxygen dissolved in
tissues at the time of irradiation as a factor in radiotherapy. Br J Radiol 1953;
26: 638–48.
36. Vaupel P, Kelleher DK, Hockel M. Oxygen status of malignant tumors: patho-
genesis of hypoxia and significance for tumor therapy. Semin Oncol 2001; 28
(Suppl 8): S29–35.
37. Dunst J, Stadler P, Becker A et al. Tumor volume and tumor hypoxia in head
and neck cancers. The amount of the hypoxic volume is important. Strahlenther
Onkol 2003; 179: 521–6.
38. Brizel DM, Dodge RK, Clough RW et al. Oxygenation of head and neck cancer:
changes during radiotherapy and impact on treatment outcome. Radiother
Oncol 1999; 53: 113–17.
39. Nordsmark M, Overgaard J. Tumor hypoxia is independent of hemoglobin and
prognostic for loco-regional tumor control after primary radiotherapy in
advanced head and neck cancer. Acta Oncol 2004; 43: 396–406.
40. Wouters BG, Weppler SA, Koritzinsky M et al. Hypoxia as a target for com-
bined modality treatments. Eur J Cancer 2002; 38: 240–57.
41. Martin L, Lartigau E, Weeger P et al. Changes in the oxygenation of head and
neck tumors during carbogen breathing. Radiother Oncol 1993; 27: 123–30.
42. Becker A, Kuhnt T, Liedtke H et al. Oxygenation measurements in head and
neck cancers during hyperbaric oxygenation. Strahlenther Onkol 2002; 178:
105–8.
43. Kaanders JH, Pop LA, Marres HA et al. Accelerated radiotherapy with
carbogen and nicotinamide (ARCON) for laryngeal cancer. Radiother Oncol
1998; 48: 115–22.
44. Kaanders JH, Pop LA, Marres HA et al. ARCON: experience in 215 patients
with advanced head-and-neck cancer. Int J Radiat Oncol Biol Phys 2002; 52:
769–78.
45. Hermans R, Lambin P, Van den Bogaert et al. Non-invasive tumor perfusion
measurement by dynamic CT: preliminary results. Radiother Oncol 1997; 44:
159–62.
46. Hermans R., Lambin P, Van der Goten A et al. Tumoral perfusion as measured
by dynamic computed tomography in head and neck carcinoma. Radiother
Oncol 1999; 53: 105–11.
47. Hermans R, Meijerink M, Van den Bogaert W et al. Tumor perfusion rate deter-
mined non-invasively by dynamic computed tomography predicts outcome in
head and neck cancer after radiotherapy. Int J Radiat Oncol Biol Phys 2003;
57: 1351–6.
48. Mullani NA, Gould KL. First-pass measurements of regional blood flow with
external detectors. J Nucl Med 1983; 24: 577–81.
49. Blomley MJK, Coulden R, Bufkin C et al. Contrast bolus dynamic computed
tomography for the measurement of solid organ perfusion. Invest Radiol 1993;
28: S72–7.
9781842143094-Ch06 8/8/07 4:24 PM Page 109
50. Blomley MJK, Coulden R, Dawson P et al. Liver perfusion studied with ultra-
fast CT. J Comput Assist Tomogr 1995; 19: 424–33.
51. Brix G, Bahner ML, Hoffmann U et al. Regional blood flow, capillary perme-
ability, and compartmental volumes: measurement with dynamic CT – initial
experience. Radiology 1999; 210: 269–76.
52. Jaschke W, Gould RG, Assimakopoulos PA et al. Flow measurements with a
high-speed computed tomography scanner. Med Phys 1987; 14: 238–43.
53. Gandhi D, Hoeffner EG, Carlos RC et al. Computed tomography perfusion of
squamous cell carcinoma of the upper aerodigestive tract. Initial results.
J Comput Assist Tomogr 2003; 27:687–93.
54. Lartigau E, Le Ridant AM, Lambin P et al. Oxygenation of head and neck
tumors. Cancer 1993; 71:2319–25.
55. Hermans R, Van den Bogaert W, Rijnders A et al. Predicting the local outcome
of glottic cancer treated by definitive radiation therapy: value of computed
tomography determined tumor parameters. Radiother Oncol 1999; 50: 39–46.
56. Mancuso AA, Mukherji SK, Schmalfuss I et al. Preradiotherapy computed
tomography as a predictor of local control in supraglottic carcinoma. J Clin
Oncol 1999; 17: 631–7.
57. Pameijer FA, Mancuso AA, Mendenhall WM et al. Evaluation of pretreatment
computed tomography as a predictor of local control in T1/T2 pyriform sinus
carcinoma treated with definitive radiotherapy. Head Neck 1998; 20: 159–68.
58. Chua DT, Sham JS, Kwong DL et al. Volumetric analysis of tumor extent in
nasopharyngeal carcinoma and correlation with treatment outcome. Int J
Radiat Oncology Biol Phys 1997; 39: 711–19.
59. Kim JH, Lee JK. Prognostic value of tumor volume in nasopharyngeal carcinoma.
Yonsei Med J 2005; 30: 221–7.
60. Chao KSC, Bosch WR, Mutic S et al. A novel approach to overcome hypoxic
tumor resistance: Cu-ATSM-guided intensity-modulated radiation therapy. Int J
Radiat Oncol Biol Phys 2001; 49: 1171–82.
61. Römer W, Muresan L, Adamietz B et al. Biological characterization and response
monitoring in head and neck cancer therapy using multislice perfusion CT.
Eur Radiol 2002; 12 (Suppl 1): S193 (abstr).
62. Million RR, Cassisi NJ, Mancuso AA et al. Management of the neck for squa-
mous cell carcinoma. In: Million RR, Cassisi NJ, eds. Management of Head
and Neck Cancer: a Multidisciplinary Approach. Philadelphia, PA: JB
Lippincott Company, 1994: 125.
9781842143094-Ch06 8/8/07 4:24 PM Page 110
9781842143094-Ch07 8/8/07 11:57 AM Page 111
7
CT perfusion applications
in lung cancer
Kwun M Fong, Rayleen V Bowman, Ian A Yang, and Kenneth A Miles
INTRODUCTION
Lung cancer is the most devastating, yet preventable, cause of prema-

ture cancer mortality and morbidity in the Western world. Smoking pre-
vention will make the most impact on lung cancer. Nonetheless,
improvements to the medical and surgical management of lung cancer
have gradually helped to refine the diagnosis, staging, and optimal
treatments, including multimodality therapy. Amongst the hurdles that
remain are the lack of effective secondary prevention, tools for effective
early diagnosis, and cost-effective new treatments aimed at molecular
targets.
Recently, it has become possible to perform dynamic contrast-
enhanced computed tomography (CT) for lung nodules and suspected
lung cancers with relatively simple modifications to standard protocols.
The intent is to capitalize on the vascular and perfusion differences that
exist between benign and malignant nodules. There are now substan-
tial data to support the notion that dynamic contrast-enhanced CT
provides quantitative information about blood flow patterns of solitary
pulmonary nodules (SPNs), and is potentially a useful diagnostic between
method for differentiating between SPNs of differing pathologies.
Indeed, it has already been utilized in a large-scale CT screening study
for lung cancer. Nonetheless, to be used in widespread clinical practice,
there needs to be standardization of an accepted protocol, and standard-
ized measurements used to ensure generalizability. Future technical
modifications are likely to make this technique even more attractive,
especially when combined with independent functional imaging
modalities, whilst taking advantage of the morphological structural and
spatial detail afforded by modern multidetector CT scanners.
9781842143094-Ch07 8/8/07 11:57 AM Page 112
PATHOLOGY OF LUNG CANCER
There are a large number of malignant tumors that affect the lungs,
ranging from metastatic malignancies to the most common type, bron-
chogenic carcinoma. Bronchogenic cancers can also metastasize to the
lungs, as well as other sites; but the lung is also a common destination
for the metastatic spread of a wide variety of other human malignancies.
Bronchogenic carcinoma (lung cancer) is the highest ranking cause
of cancer-associated mortality in the world. Historically more prevalent
in men, bronchogenic carcinoma is currently causing higher mortality
among women living in several Western countries than is breast or any
other cancer. Non-small-cell lung cancer accounts for approximately
80–85% of primary lung cancer, and small-cell cancer accounts for most
of the remainder. The two types differ in both biological characteristics
and clinical behavior.
Presentation
Lung cancers can present in a large number of ways. A frequently
encountered clinical situation is the presentation of a patient with or
without symptoms who is found to have a pulmonary nodule on chest
radiography or on thoracic CT undertaken for another reason, such as
diagnosis of suspected pulmonary embolism. For this common clinical
scenario where lung cancer is a diagnostic possibility, the challenge is
first to ascertain the presence or absence of malignancy. If malignancy
is confirmed, the clinician has to determine the origin of the cancer, and
finally the extent or stage of disease so that therapy may be appropri-
ately individualized. Therefore, tools to help decide the likelihood of
malignancy and the subsequent need for invasive investigations for
pulmonary lesions are essential for the clinician who manages patients
with suspected lung cancer.
Increasing detection of pulmonary nodules

in clinical practice
The increasingly frequent use of sensitive techniques such as CT scan-
ning has led to more recognition of pulmonary nodules. Low-dose hel-
ical CT scanning is several times more sensitive than conventional chest
radiography for detecting pulmonary nodules, and could come into
routine practice as a screening tool if current randomized trial data
show it to be effective in reducing lung cancer mortality. The potential
impact of these sensitive imaging tools on clinical practice is illustrated
9781842143094-Ch07 8/8/07 11:57 AM Page 113
CT perfusion applications in lung cancer 113
by large screening studies in some parts of the world, which show non-
calcified nodules in the majority of screenees.1 Therefore, if low-dose
CT comes into routine use, there will be an exponential increase in the
number of nodules that require clinical evaluation.
As most identified pulmonary nodules (defined as an intraparenchy-
mal lung lesion less than 3 cm in diameter and not associated with atelec-
tasis or lymphadenopthy2 turn out not to be malignant, there is a pressing
need for strategies to efficiently triage pulmonary nodules so that those
requiring treatment can be identified, while avoiding the expense, risk,
and inconvenience of unnecessary invasive investigations.
CONTRAST-ENHANCED CT
CT measurement of nodule enhancement with iodinated contrast media

is an example of so-called dynamic or functional imaging. The physio-
logical basis is the difference in vascularity and vasculature of neoplastic
compared to benign nodules.3 Investigators have exploited these differ-
ences with a number of distinct modalities to show that lung cancers
enhance more than benign lung lesions. Apart from contrast-enhanced
CT, differences in tumor vascularity have been demonstrated with
angiography,4 contrast-enhanced tomography,5 fluorodeoxyglucose-
positron emission tomography (FDG-PET),6 Doppler ultrasound,7 and
magnetic resonance imaging.8
TECHNICAL CONSIDERATIONS: MEASUREMENT OF

CONTRAST ENHANCEMENT AND PERFUSION
With the advent of spiral CT, there has been a rapid profusion of
research protocols aimed at distinguishing differences in vascularity
between malignant and benign tumors. In addition, there have also
been a variety of measurements utilized for quantifying the results from
contrast-enhanced nodule scanning.
Simple measurements of tissue attenuation9 or enhancement10 can be
obtained, as can absolute measures of tissue perfusion.11 Peak enhance-
ment is a measure of the maximum increase in density after contrast
administration, and is determined by not only tumor perfusion but also
blood volume and capillary permeability. The degree to which peak
enhancement approximates perfusion is dependent upon the transit time
of contrast material through the tumor circulation: the longer is the tran-
sit time, the closer is the dependence of peak enhancement on perfusion.
9781842143094-Ch07 8/8/07 11:57 AM Page 114
Transit time is in turn affected by tissue blood volume and vascular

permeability. All three parameters, perfusion, blood volume, and per-
meability, tend to be increased in malignancy, resulting in greater
enhancement. The use of peak enhancement measures can be
extended to include an evaluation of the washout of contrast material
by incorporating an additional acquisition at 15 minutes.12
A typical protocol for measuring pulmonary nodule enhancement
from a series of spiral acquisitions is illustrated in Box 7.1.13 Enhancement
can be determined by subtracting the baseline density value in a region
of interest (ROI) constructed within the nodule from the equivalent
value in the ROI on an image obtained at peak contrast enhancement
(Figure 7.1a). A number of factors must be considered, to enable such
techniques to produce consistent results. First, the degree of enhance-
ment for a given dose of iodine is dependent upon weight, and thus
Box 7.1 Technique for quantifying contrast enhancement within lung nodules: repeated
spiral (helical) acquisitions (based on reference 13)
Contraindications: Usual contraindications for contrast-enhanced CT.

System calibration (preferable but not essential):
1. Position a phantom containing tubes of contrast material (300–370 mg/ml) diluted in
normal saline at ratios of 1:20, 1:35, 1:50, 1:100, and 1:200 so that the tubes are aligned
along the central z-axis of the CT system.
2. Acquire and reconstruct images as for the patient protocols below.
3. On reconstructed images of the phantom, place regions of interest (ROIs) over each
tube containing contrast material.
4. Plot attenuation measurements from each tube against the corresponding iodine
concentration and determine the iodine calibration factor from the gradient of a linear
least-squares fit of the five points.
Image acquisition:
1. Pre-contrast-enhancement images to locate the nodule: 3 mm collimation, pitch 1:1,
120 kVp, 280 mA.
2. Contrast-enhanced images; z-axis cluster of at least 15 mm acquired at 1, 2, 3, and
4 minutes following 420 mg iodine/kg injected at 600 mg/minute (e.g. 300 mg/ml at
2 ml/s); 3 mm collimation, pitch 1:1, 120 kVp, 280 mA. Suspended shallow inspiration.
Image analysis:
1. Reconstruct images at 2 mm intervals.
2. Place a circular or oval ROI within the nodule as displayed on mediastinal windows,
ROI size approximately 70% of nodule long and short axis diameters.
3. Determine enhancement by subtracing baseline nodule attenuation from attenuation
values on subsequent enhanced images.
4. If system calibration performed, use calibration factor to convert enhancement to
mg iodine.
Interpretation: Enhancement greater than 15 HU (approximately 0.55 mg iodine/ml)
indicates possible malignancy.
9781842143094-Ch07 8/8/07 11:57 AM Page 115
a b c
Enhancement = 30 HU
Figure 7.1 Assessment of likely malignancy within a solitary pulmonary

nodule. (a) Measurement of contrast enhancement. (b) Parametric image
of contrast enhancement image to evaluate irregularity of enhancement.
(c) Corresponding fluorodeoxyglucose-positron emission tomography (FDG-PET)
image.
contrast material should be administered on a dose-by-weight basis.

Alternatively, enhancement values should be corrected for body weight.
Second, most studies adopting this methodology fail to consider the
necessity for calibrating the CT system for sensitivity to iodine. It is well
recognized that reducing the tube voltage will increase the sensitivity of
CT systems to iodine. However, it is less well appreciated that, despite
routine quality control measures, significant differences in iodine sensi-
tivity can exist between different CT systems, and that the sensitivity of
a single CT system can change significantly with time.14 Hence, 15 HU
of enhancement on one CT system could be equivalent to 17.6 HU on
the same system 45 weeks later, or to 18.7 HU on another system.14
Regular calibration of CT systems using a purpose-built phantom
(Figure 7.2) would overcome this problem of variability, and enable
contrast enhancement to be more accurately expressed in terms of
iodine concentration. Even with these corrections, enhancement
measures remain dependent upon cardiac output.11
Parametric images of peak enhancement can also be generated to allow
an assessment of enhancement heterogeneity (Figure 7.1b). Malignant
nodules are more likely to exhibit irregular enhancement. Similarly, it is
also possible to calibrate such images so that they are quantified as the
standardized enhancement value (SEV) or standardized perfusion value
(SPV) (Box 7.2). For the SEV, enhancement values are normalized to
the enhancement expected if contrast is evenly diluted within the
patient volume.11 Similarly, the SPV relates tissue perfusion to average
whole-body perfusion, thereby allowing comparison of various perfu-
sion measurements.11 The derivation of these parameters is analogous
to the standardized uptake value (SUV) used in FDG-PET, and allows
9781842143094-Ch07 8/8/07 11:57 AM Page 116
a
T
1 5
2 4
DATA SPECTRUM CORPORATION V
Figure 7.2 Phantom for calibrating computed tomography (CT) systems for
sensitivity to iodine. (a) End plan-view and (b) overall view. Contrast material
at various dilutions is introduced into cylinders of diameter 25 mm and length
50 mm located at positions 1–5 as follows: (1) 1:50, (2) 1:35, (3) 1:20,
(4) 1:200, (5) 1:100. Three small venting plugs (V) allow release of air when filling
the phantom with water. (Reproduced with permission from reference 14)
more direct comparison of the results of the two modalities.11 Generation

of such images is aided by the selection of appropriate thresholds to
subtract normal lung tissue and blood vessels (see Chapter 4). By using
volumetric acquisitions, it is possible to correct for respiratory motion
by selecting corresponding images from the baseline and peak
enhancement datasets.
Alternatively, perfusion of the nodule can be calculated from the
density–time curve of the tissue of interest adjusted for arterial input
obtained from the density–time curve within the thoracic aorta. Perfusion
measurements are less affected by contrast dose, patient weight, and
cardiac output, but require dedicated image acquisition and more
complex calculations.11 However, image misregistration due to respira-
tory motion and beam hardening artifacts from high-density contrast
9781842143094-Ch07 8/8/07 11:57 AM Page 117
Box 7.2 Protocol for generating consistent CT images of peak enhancement within
lung masses
Aim: To generate images of peak enhancement within lung nodules whilst minimizing
inconsistencies due to motion and/or segmentation of the tumor. The resulting images
will allow assessment of regional variations in tumor peak enhancement.
1. Create a region of interest (ROI) encompassing the pulmonary nodule displayed with
mediastinal windows.
2. Use image thresholds of −50 HU to 100 HU to eliminate from analysis any pixels
containing air or calcium.
3. Exclude large blood vessels from the ROI by eliminating pixels that are enhanced by
more than 100 HU following contrast material.
4. Display the changes in X-ray attenuation within the ROI following administration of
contrast material as a time-attenuation curve and identify the time of maximal tumor
enhancement. Only the image at this time point and the baseline image are retained
for further analysis.
5. If several images have been acquired at each time, compensate for respiratory motion
by selecting from each data set the images that correspond most closely in anatomical
position.
6. Generate parametric image displaying peak tumor enhancement (and/or standardized
perfusion value) by subtracting the baseline image from the image in which the
attenuation within tumor ROI was maximal.
material in the venous system can significantly impair perfusion images

of lung lesions.11 Furthermore, lung tumors may be supplied by either
the pulmonary or the bronchial circulation. Thus, when using deconvo-
lution analysis, the choice of blood vessel for the input function is an
important consideration when undertaking CT perfusion of lung lesions,
because significant errors can occur if an inappropriate input vessel is
selected.
EVALUATION OF PULMONARY NODULES
Based on the differences in blood flow between benign and malignant

nodules, contrast-enhanced CT has been proposed as a useful tool for
evaluating indeterminate pulmonary nodules.15
Prior to this, Littleton et al.5 had demonstrated the potential for using
contrast enhancements to aid in the diagnosis of lung cancer in 45 patients
with pulmonary masses using conventional tri-spiral tomograms. Their
results suggested that contrast enhancement of lung masses could be
measured on sectional images, and set the scene for exploiting the
power of CT imaging.
9781842143094-Ch07 8/8/07 11:57 AM Page 118
Improving on this time-consuming densitometric measurement of

film tomograms, Swensen and colleagues initially reported a preliminary
study of a contrast-enhanced thin-collimation CT technique in evaluat-
ing solitary pulmonary nodules in 52 patients.16 All malignant nodules
had enhanced by 20 HU or more within the first 2 minutes after
contrast injection, compared with only one benign nodule. Since then,
multiple subsequent investigations of the technique in evaluating
pulmonary nodules have been reported (Table 7.1).
Overall, more malignant tumors enhance following contrast material
(median 46.5 HU, range 11–110 HU) compared with granulomas and
benign lesions.10 Setting 20 HU as the threshold for a positive test, this
prospective study reported by Swensen et al. found that the sensitivity
was 98% and the specificity 73%. From another perspective, enhance-
ment of less than 15 HU in a nodule appeared to be strongly predictive
of benignity. In a subset of 24 nodules, CT contrast enhancement was
found to correlate with the degree of central immunohistochemical
staining for an antibody to factor VIII-associated antigen.
In a different population of 32 solitary pulmonary nodules, includ-
ing lung cancer, tuberculomas, and hamartomas, Yamashita et al. found
that all lung cancers, but only one benign lesion, a hamartoma, showed
Table 7.1 Summary of studies using contrast enhancement to evaluate solitary

pulmonary nodules
No. of Sensitivity Specificity

Study patients Technique Criteria (%) (%)
Swensen et al.13 356 Multispiral > 15 HU 98 58

Yamashita et al.9 32 Multispiral 20–60 HU 88 36
Zhang and Kono17 65 Single > 20 HU 95 85
location > 0.2 ml/min
Miles et al.11 11 Single SPV ≥ 1.6 100 50
location
Petkovska et al.18 29 Multispiral > 15 HU 93 21
Jeong et al.12 107 Multispiral ≥ 25 HU and 94 90
washout
3–31 HU
(> 25 HU only) (100) (48)
Yi et al.19 119 Multispiral ≥ 25 HU and 81 93
washout
3–31 HU
(> 25 HU only) (95) (60)
Christensen et al.20 42 Multispiral > 15 HU 100 29
SPV, standardized perfusion value

9781842143094-Ch07 8/8/07 11:57 AM Page 119
CT contrast enhancement using a protocol that scanned before, and

30 seconds, 2 minutes, and 5 minutes after contrast material administra-
tion.9 The time–attenuation curve and the maximum attenuation of a
nodule were considered to reflect the vascularity and blood pool of the
nodule respectively. Overall, the time–attenuation curve for the 18 lung
cancers showed gradual enhancement, whereas the maximum attenua-
tion for these cancers ranged from 25 to 56 HU.
Similar findings of peak enhancement after contrast CT in malignant
(42 HU) and inflammatory benign (44 HU) nodules compared to non-
inflammatory nodules (13 HU) in 65 patients were reported by Zhang and
Kono.17 These investigators also measured the peak height of time–atten-
uation curves and ratio of the peak height of the SPN to that of the aorta.
SPN-to-aorta ratios in malignant and inflammatory SPNs were significantly
higher than in benign SPNs. Perfusion was additionally calculated from
the maximum gradient of the time–attenuation curve and the peak height
of the aorta. Again, perfusion values in malignant and inflammatory SPNs
were significantly higher than those of the benign SPNs. However, no
statistically significant differences in the peak height and SPN-to-aorta ratio
were found between malignant and inflammatory SPNs.
To test the hypothesis that the absence of nodule enhancement, that
is 15 HU or less, is predictive of benignity, a large multicenter study of lung
nodule CT enhancement was performed by Swensen and his colleagues.
Three hundred and fifty-six eligible nodules from an initial 550 patients
from seven centres were studied for peak nodule enhancement, and
time–attenuation curves.13 As expected, malignant lesions enhanced
(median 38.1 HU, range 14.0–165.3 HU) significantly more than granu-
lomas and benign lesions (median 10.0 HU, range −20.0–96.0 HU). In
this study, using 15 HU as the threshold, sensitivity was again very high
at 98% (167 of 171 malignant nodules) with a specificity of 58% (107 of
185 benign nodules), thus supporting the hypothesis that contrast
enhancement of 15 HU or less predicts a benign lesion.
Similar results have been obtained in two recent comparisons of
nodule enhancement versus FDG-PET. Using 15 HU of enhancement as
the diagnostic threshold, Christensen et al.20 reported a sensitivity and
specificity for nodule enhancement of 100% and 29%, respectively,
whereas Yi et al.19 reported a sensitivity of 95% and specificity of 60%
when using net enhancement of greater than or equal to 25 HU.
CT enhancement and pathology of lung nodules

In a separate radiologic–pathologic comparison of contrast-enhanced
CT in small peripheral lung cancers, Yamashita et al. reported that
9781842143094-Ch07 8/8/07 11:57 AM Page 120
malignant lung carcinomas had maximum attenuation (difference in CT

numbers between unenhanced attenuation and maximum enhancement)
of between 20 and 89 HU.21 They demonstrated that enhancement cor-
related with the number of small vessels, and the distribution of elastic
fibers in the interstitium.
A study of histologically confirmed lung cancers found no significant
differences in peak enhancement or perfusion between small-cell lung
cancer and non-small-cell lung cancer.22 Perfusion of larger tumors was
lower than that of smaller tumors, as was the case for central tumors
compared to peripheral tumors. On the other hand, peak enhancement
of central and peripheral cancers did not differ significantly. By study-
ing the enhancement characteristics of the cancer compared with the
aorta, it was also concluded that some tumors were supplied by both
pulmonary and/or bronchial vessels.
More recently, Yi and colleagues reported the correlation of dynamic
enhanced helical CT and vascular endothelial growth factor (VEGF)
immunostaining and microvessel density.23 Using a sensitivity of
10 HU or more for designating malignant nodules, this protocol resulted
in a very high sensitivity of 99% (69 of 70) for malignant nodules but a
lower specificity of 54% (33 of 61 benign modules), from 131 patients
with solitary pulmonary nodules. The protocol involved nine series of
images obtained at 20-second intervals for 3 minutes after contrast
medium injection. Peak enhancement but not net enhancement was
found to positively correlate with microvessel counting using Weidner’s
method, after CD31 staining and VEGF immunostaining with a poly-
clonal antibody.23
The potential impact of intratumoral necrosis on contrast enhance-
ment has been studied by Yamashita et al.24 One of four peripheral lung
cancers exhibiting non-cavitary necrosis within the surgically resected
specimen was associated with poor enhancement, highlighting this
pathological feature as a potential pitfall in the use of nodule enhance-
ment for characterizing lung nodules.
Comparison with FDG-PET scans

Several groups have sought to establish the relation between contrast-
enhanced CT and FDG-PET in lung tumors. In 40 consecutive lung can-
cers, Tateishi et al. found that contrast enhancement (as measured by
peak attenuation) and relative flow (from dynamic contrast-enhanced
CT) correlated with the standardized uptake value (SUV) of FDG-PET
as well as intratumoral microvessel density revealed by immunoreactiv-
ity to a CD34 antibody.25 A study comparing FDG-PET SUV and SPV
9781842143094-Ch07 8/8/07 11:57 AM Page 121
from contrast-enhanced CT in 18 non-small-cell lung cancers showed

that the ratio of SUV to SPV and the metabolic–blood flow difference
(SUV−SPV) were not altogether consistent, but appeared to vary
according to tumor size and stage.26 The apparent uncoupling of tumor
blood flow and metabolism in larger tumors can be postulated to reflect
a possible metabolic adaptation of larger cancers which have outgrown
their blood supply.
Two direct comparisons of quantitative contrast-enhanced CT and
PET have shown superior diagnostic accuracy for FDG-PET in the char-
acterization of pulmonary nodules.19,20 However, using enhancement
criteria alone, each study found the sensitivity, and hence negative pre-
dictive value, to be comparable for the two techniques. Thus, in view of
the greater availability and lower cost of contrast-enhanced CT, it is pos-
sible to envisage a diagnostic strategy which incorporates nodule con-
trast-enhancement in the selection of patients for FDG-PET (Figure 7.3).
Nodules which demonstrate low enhancement could be watched, whilst
nodules that enhance would undergo FDG-PET. A decision tree analysis
has outlined the potential cost-effectiveness of this approach.27 In the
Australian setting, this strategy can be estimated to save $117 per
patient compared to investigation with conventional CT with PET, and
Solitary Thin Benign

pulmonary section features
nodule CT
Indeterminate
Nodule 2 year CT
Low
enhanced follow-up for
enhancement
CT stability
High enhancement
FDG Negative
PET
Positive
Biopsy/
resection
Figure 7.3 Strategy for the cost-effective investigation of pulmonary nodules

with conventional CT, nodule-enhanced CT, and FDG-PET.
9781842143094-Ch07 8/8/07 11:57 AM Page 122
would remain more cost-effective provided that the prior probability of

malignancy is less than 80%.
POTENTIAL LIMITATIONS
Variations in technique
To date, there does not appear to be any consensus as to a standard-
ized protocol, including the amount of contrast material and flow rate
to be used, the frequency of data recording, the scanning time, and
the scan interval. Each of these factors has the potential to affect the
pattern and quantity of measurable enhancement.28
Enhancing benign lesions

Evidence including that from a multicenter prospective study indicates
that the absence of contrast enhancement predicts the benign etiology
of a pulmonary nodule. In contrast, not all enhancing nodules are
malignant. Some are inflammatory nodules, and others are intrapul-
monary lymph nodes.29,30
Small nodules
Most if not all studies report contrast enhancement in pulmonary nodules
of at least 5 or 6 mm in diameter.3,16,23 There is limited knowledge of the
value of this technique in very small lesions.
Tumoral heterogeneity
Yamashita et al. noted that some lung cancers demonstrate inhomo-
geneous patterns of contrast enhancement at CT, some correlating
to necrosis, fibrosis, or perhaps differentiation.21 This implies that
necrotic cancers may not demonstrate the high levels of enhancement
seen with non-necrotic tumors (Figure 7.4).24 Indeed some experts
recommend that the use of contrast-enhanced CT be restricted to
pulmonary nodules with diameters 2–2.5 cm or less, as they are less
likely to have substantial necrosis, and thus be falsely negative.3,26
Furthermore, the clinical impact of smaller lesions is different, as they
are more likely to be benign, and are generally more difficult to
biopsy successfully either bronchoscopically or via a transthoracic
percutaneous approach.
9781842143094-Ch07 8/8/07 11:57 AM Page 123
a b 50HU
0HU
Figure 7.4 Large non-small-cell lung cancer with low-level and heterogeneous
enhancement implying necrosis. (a) Conventional contrast-enhanced CT;
(b) maximum-enhancement image.
Potential technical artifacts

Potential factors that may result in artifacts include beam hardening
artifact in lesions localized directly to large central vessels, and motion
artifact from longer breath-holds. For instance, in the large multicenter
study, a minority of nodules were reported as technically inadequate
because of respiratory misregistration.13
Radiation dose
Many contrast CT scans are utilizing tube voltages of 80 kV in an attempt
to reduce radiation exposure. This voltage also has the advantage of
increasing the sensitivity to iodine-based contrast agents. However, for
measures of nodule enhancement, new diagnostic thresholds will need
to be established for reduced tube voltages. It has been estimated using
International Commission on Radiation Protection (ICRP) recommenda-
tions and the CT-Expo program for CT dose evaluation that there is an
effective dose of about 1.3 mSv.22
FUTURE DEVELOPMENTS
Refinements
Heterogeneous enhancement is more commonly a feature of malignant
nodules.18 Although heterogeneity can be assessed visually, this image
9781842143094-Ch07 8/8/07 11:57 AM Page 124
feature is also amenable to quantification through computer analysis.

With the exponential increase in computing power, such future refine-
ments are likely to be rapidly applied to this technique by various
investigators. Preliminary data indicate that computer-aided diagnosis
(CAD) appears to be able to improve the determination of malignant
from benign nodules based on quantitative features extracted from
volumetric thin-section CT image data acquired before and after the
injection of contrast medium.31 For instance, it is possible that visual
semiquantitative and quantitative characterization of contrast enhance-
ment patterns may potentially improve the discrimination between
benign and malignant nodules.18
Another group has suggested that analyzing combined wash-in and
washout contrast-enhanced characteristics with multidetector row CT
can improve the specificity of contrast-enhanced CT from 48% to 90%
with only a slight fall in sensitivity from 100% to 94%.12 Similar results
were found when using the same criteria in a subsequent comparison
with FDG-PET (sensitivity: 81%, specificity: 93%).19
Prognostication and treatment monitoring

A meta-analysis has confirmed that tumor microvessel count is a poor
prognostic factor for survival in surgically treated non-small-cell lung
cancer.32 As contrast enhancement correlates with tumor microvessel
count,22 it is possible that contrast enhancement may prove to be a
useful preoperative prognostic factor. Indeed, Shim et al. have shown
that higher enhancement in non-small-cell lung cancer is predictive of
advanced nodal status (see Chapter 10).33 However, this hypothesis
requires further testing in prospective clinical trials.
Contrast CT has the potential to help predict and/or monitor the
response to treatment (see Chapter 12). As proof of principle, Kiessling
et al. demonstrated a reduction in tumor perfusion following
chemotherapy.22 Whilst this may potentially help to distinguish between
residual tumor and post-treatment scar, it remains to be tested and val-
idated in appropriately designed and powered prospective studies. Choi
et al. reported that the treatment of small-cell lung cancer also resulted
in lower enhancement, but found that the pretreatment enhancement
level could predict the likely response.34
Whole tumor quantitation

Until recently, contrast-enhanced CT measurements have been performed
with multidetector row CT limited to a single tumor level comprising
9781842143094-Ch07 8/8/07 11:57 AM Page 125
contiguous transverse sections (maximum coverage of 4 cm using

64-detector row scanners).35 This means that spatial heterogeneity
within the tumor vasculature may render CT perfusion measurements
prone to potential misrepresentation. To overcome this, a novel and
reproducible helical acquisition technique has recently been described,
utilizing Patlak analysis, which allows the vascular status of the entire
tumor to be analyzed.35 If further validated, this improvement has the
potential to reliably assess whole tumors rather than arbitrary single
tumor sections.
CONCLUSIONS
Conventional CT imaging is now a routine diagnostic study for evaluat-

ing suspected lung cancers and pulmonary nodules, staging, and
following lung cancers after treatment. There have been significant
advances in the field of functional imaging including techniques such
as FDG-PET scans, contrast-enhanced CT, and magnetic resonance
imaging (MRI).8
Contrast-enhanced CT offers improved accuracy in the prediction of
malignancy in lung nodules, particularly when combined with morpho-
logical analysis. There are variations to the technique, but in general the
results are generally concordant between techniques. Furthermore,
some concordance with alternative, independent imaging methods for
demonstrating tumor blood flow has been demonstrated.
Already, contrast-enhanced CT has been used in a large-scale,
low-dose CT screening study.36 A study of repeated yearly spiral CT and
selective use of PET in a large cohort of high-risk volunteers enrolled
1035 individuals aged 50 years or older who had smoked for 20 pack-
years or more. All subjects underwent annual low-dose CT, with or with-
out PET, for 5 years. Lesions up to 5 mm were deemed non-suspicious,
and low-dose CT was repeated after 12 months (year 2). Spiral thin-section
CT limited to the area of interest and three-dimensional analysis were
used for non-calcified lesions greater than 5 mm, with assessment of
contrast enhancement in nodular lesions that had a density of more than
0 HU. Positively enhancing (threshold of > 30 HU), PET-positive, and
non-calcified uncertain lesions of 20 mm or larger were recommended
for biopsy.
The development of standardized measures and terminology for
enhancement and perfusion should allow cross-comparison between
studies and, more important, across diagnostic centers. CT has some
advantages compared to other techniques, including broad availability
9781842143094-Ch07 8/8/07 11:57 AM Page 126
compared to MRI or PET, worldwide expertise and familiarity with CT,

and relative simplicity and cost. Moreover, there is some evidence from
decision tree analysis that contrast-enhanced CT may be a cost-effective
approach for the evaluation of solitary pulmonary nodules.27 The chal-
lenge is to agree on a simple universally accepted protocol, to test this
protocol in large-scale studies to confirm its discriminatory power for
malignant and benign nodules, and to implement its use for the day to
day evaluation of suspicious pulmonary nodules.
In the future, there will be much interest in exploiting the pattern of
contrast enhancement in combined PET–CT examinations, in addition to
the accepted benefit delivered by CT, such as anatomic correlation and
attenuation correction for PET.37 The potential of the technique is likely to
be magnified in the future by advanced computer software and hardware
developments. Moreover, it may be that dynamic contrast-enhanced CT
will become a useful adaptation of conventional fusion PET–CT imaging
for the assessment of lung nodules and lung cancers. There may also be
a therapeutic use in the characterization of lung cancer vascularity to pre-
dict the response of tumors to the new vascular targeting and antiangio-
genic agents that are reaching the stage of large clinical trials, especially if
whole-tumor measurements can be simply obtained.
REFERENCES
1. Swensen SJ, Jett JR, Sloan JA et al. Screening for lung cancer with low-dose
spiral computed tomography. Am J Respir Crit Care Med 2002; 165: 508–13.
2. Tuddenham WJ. Glossary of terms for thoracic radiology: recommendations of
the Nomenclature Committee of the Fleischner Society. AJR Am J Roentgenol
1984; 143: 509–17.
3. Swensen SJ. Functional CT: lung nodule evaluation. Radiographics 2000; 20:
1178–81.
4. Viamonte M Jr. Angiographic evaluation of lung neoplasms. Radiol Clin North
Am 1965; 3: 529–42.
5. Littleton JT, Durizch ML, Moeller G, Herbert DE. Pulmonary masses: contrast
enhancement. Radiology 1990; 177: 861–71.
6. Patz EF Jr, Lowe VJ, Hoffman JM et al. Focal pulmonary abnormalities:
evaluation with F-18 fluorodeoxyglucose PET scanning. Radiology 1993; 188:
487–90.
7. Yuan A, Chang DB, Yu CJ et al. Color Doppler sonography of benign and
malignant pulmonary masses. AJR Am J Roentgenol 1994; 163: 545–9.
8. Guckel C, Schnabel K, Deimling M, Steinbrich W. Solitary pulmonary nodules:
MR evaluation of enhancement patterns with contrast-enhanced dynamic
snapshot gradient-echo imaging. Radiology 1996; 200: 681–6.
9. Yamashita K, Matsunobe S, Tsuda T et al. Solitary pulmonary nodule: prelimi-
nary study of evaluation with incremental dynamic CT. Radiology 1995; 194:
399–405.
9781842143094-Ch07 8/8/07 11:57 AM Page 127
10. Swensen SJ, Brown LR, Colby TV, Weaver AL, Midthun DE. Lung nodule
enhancement at CT: prospective findings. Radiology 1996; 201: 447–55.
Radiology 2001; 220: 548–53.
12. Jeong YJ, Lee KS, Jeong SY et al. Solitary pulmonary nodule: characterization
with combined wash-in and washout features at dynamic multi-detector row
CT. Radiology 2005; 237: 675–83.
919–26.
15. Erasmus JJ, Connolly JE, McAdams HP, Roggli VL. Solitary pulmonary nodules:
Part I. Morphologic evaluation for differentiation of benign and malignant
lesions. Radiographics 2000; 20: 43–58.
16. Swensen SJ, Morin RL, Schueler BA et al. Solitary pulmonary nodule: CT eval-
uation of enhancement with iodinated contrast material—a preliminary report.
Radiology 1992; 182: 343–7.
17. Zhang M, Kono M. Solitary pulmonary nodules: evaluation of blood flow
patterns with dynamic CT. Radiology 1997; 205: 471–8.
18. Petkovska I, Shah SK, McNitt-Gray MF et al. Pulmonary nodule characterization:
a comparison of conventional with quantitative and visual semi-quantitative
analyses using contrast enhancement maps. Eur J Radiol 2006; 59: 244–52.
19. Yi CA, Lee KS, Kim BT et al. Tissue characterization of solitary pulmonary
nodule: comparative study between helical dynamic CT and integrated
PET/CT. J Nucl Med 2006; 47: 443–50.
20. Christensen JA, Nathan MA, Mullan BP et al. Characterization of the solitary
pulmonary nodule: 18F-FDG PET versus nodule-enhancement CT. AJR Am J
Roentgenol 2006; 187: 1361–7.
21. Yamashita K, Matsunobe S, Takahashi R et al. Small peripheral lung carcinoma
evaluated with incremental dynamic CT: radiologic-pathologic correlation.
Radiology 1995; 196: 401–8.
22. Kiessling F, Boese J, Corvinus C et al. Perfusion CT in patients with advanced
bronchial carcinomas: a novel chance for characterization and treatment
monitoring? Eur Radiol 2004; 14: 1226–33.
23. Yi CA, Lee KS, Kim EA et al. Solitary pulmonary nodules: dynamic enhanced
multi-detector row CT study and comparison with vascular endothelial growth
factor and microvessel density. Radiology 2004; 233: 191–9.
24. Yamashita K, Matsunobe S, Tsuda T et al. Intratumoral necrosis of lung
carcinoma: a potential diagnostic pitfall in incremental dynamic computed
tomography analysis of solitary pulmonary nodules? J Thorac Imaging 1997;
12: 181–7.
25. Tateishi U, Nishihara H, Tsukamoto E et al. Lung tumors evaluated with FDG-
PET and dynamic CT: the relationship between vascular density and glucose
metabolism. J Comput Assist Tomogr 2002; 26: 185–90.
26. Miles KA, Griffiths MR, Keith CJ. Blood flow-metabolic relationships are depend-
ent on tumour size in non-small cell lung cancer: a study using quantitative
contrast-enhanced computer tomography and positron emission tomography.
Eur J Nucl Med Mol Imaging 2006; 33: 22–8.
9781842143094-Ch07 8/8/07 11:57 AM Page 128
27. Comber LA, Keith CJ, Griffiths M, Miles KA. Solitary pulmonary nodules:
impact of quantitative contrast-enhanced CT on the cost-effectiveness of
FDG-PET. Clin Radiol 2003; 58: 706–11.
28. Marten K, Grabbe E. The challenge of the solitary pulmonary nodule: diagnostic
assessment with multislice spiral CT. Clin Imaging 2003; 27: 156–61.
29. Diederich S, Theegarten D, Stamatis G, Luthen R. Solitary pulmonary nodule
with growth and contrast enhancement at CT: inflammatory pseudotumour as
an unusual benign cause. Br J Radiol 2006; 79: 76–8.
30. Matsuki M, Noma S, Kuroda Y et al. Thin-section CT features of intrapul-
monary lymph nodes. J Comput Assist Tomogr 2001; 25: 753–6.
31. Shah SK, McNitt-Gray MF, Rogers SR et al. Computer aided characterization of
the solitary pulmonary nodule using volumetric and contrast enhancement
features. Acad Radiol 2005; 12: 1310–19.
32. Meert AP, Paesmans M, Martin B et al. The role of microvessel density on the
survival of patients with lung cancer: a systematic review of the literature with
meta-analysis. Br J Cancer 2002; 87: 694–701.
cancer enable the prediction of hilar or mediastinal nodal metastasis? AJR Am
J Roentgenol 2006; 186: 981–8.
34. Choi J-B, Park C-K, Park DW et al. Does contrast enhancement on CT suggest
tumor response for chemotherapy in small cell carcinoma of the lung?
J Comput Assist Tomogr 2002; 26: 797–800.
35. Ng QS, Goh V, Fichte H et al. Lung cancer perfusion at multi-detector row CT:
reproducibility of whole tumor quantitative measurements. Radiology 2006;
239: 547–53.
36. Pastorino U, Bellomi M, Landoni C et al. Early lung-cancer detection with
spiral CT and positron emission tomography in heavy smokers: 2-year results.
Lancet 2003; 362: 593–7.
37. Antoch G, Stattaus J, Nemat AT et al. Non-small cell lung cancer: dual-modality
PET/CT in preoperative staging. Radiology 2003; 229: 526–33.
9781842143094-Ch08 8/8/07 12:02 PM Page 129
8
Tumors of the
gastro-intestinal tract
Part A: Rectal cancer

Massimo Bellomi and Giuseppe Petralia
BACKGROUND
Although screening programs and recent advances in rectal cancer

treatment such as total mesorectal excision (TME) and neoadjuvant
chemoradiation therapy, have improved survival both in Europe and in
the United States,1 the American Cancer Society estimates that there will
be about 40 340 new cases of rectal cancer in 2005 in the United States,2
and only 50% of these patients will be alive in 2010.3 Innovative treat-
ments are required to improve these outcomes. Some researchers have
selected angiogenesis as a promising target for novel therapies4–6 for
rectal cancer, since high values of tumor microvessel density (MVD)
have been correlated with poor outcome,7,8 and high vascular endo-
thelial growth factor (VEGF) expression in this malignancy has been
associated with disease progression and poor survival.9,10
An in-vivo marker of rectal cancer angiogenesis is, then, required for
therapeutic monitoring of new antiangiogenic trials, since morphologic
imaging, based on size criteria, may not be suitable for monitoring the
effects of antiangiogenesis drugs.11
Furthermore, angiogenesis assessment may be extremely useful
for diagnosis, risk stratification, and conventional chemo- and radiation
therapy monitoring in rectal cancer patients.
Computed Tomography Perfusion imaging (CTP) has shown potential
as a functional imaging tool for angiogenesis assessment.12–15 It is an easy
examination, which can be performed with conventional scanners and
9781842143094-Ch08 8/8/07 12:02 PM Page 130
conventional contrast medium using commercially available software,

and it has shown encouraging results in rectal cancer assessment.16–18
TECHNIQUE
CTP of rectal cancer is a potentially time-consuming examination if

standardized protocols for patient preparation, image acquisition, and
contrast medium administration are not observed. In the authors’ expe-
rience, the estimated time for CTP of the rectum, from patient prepara-
tion to discharge, is 12–15 minutes, if protocols are strictly observed.
Patient preparation
Adequate bowel preparation is required for accurate rectal cancer imaging:
24 h before the CTP examination a diet with low amounts of fiber is
observed and a cleansing enema is performed, combined with oral
laxative administration.
No oral contrast is administered to the patient. Once on the com-
puted tomography (CT) bed, a 24-Fr Foley catheter is positioned as low
as possible in the rectum, positioning the inflated balloon against the
inner margin of the anal sphincter, and a 2-liter water enema is admin-
istered for homogeneous, reproducible and persistent rectal wall disten-
tion and visualization. Non-contrast-enhanced CT of the pelvis (2.5 mm
slice thickness) is performed to locate the rectal cancer.
One milliliter of hyoscin butylbromide is administered immediately
before contrast medium injection and dynamic scanning, in order to
avoid movement artifacts due to bowel peristalsis.
Image acquisition protocol

A comprehensive CTP study, including both first-pass and delayed
imaging, may be performed for studying rectal cancer, since the rectum
is not affected by respiratory or other relevant motions. The matrix
detector coverage, using a 16-slice GE CT (LightSpeed®; General Electric
Medical Systems, Milwaukee, WI, USA), is 20 mm, and in cine mode it
can be used to obtain 4 × 5-mm or 2 × 10-mm slices. Slices are selected
to include the maximum visible tumor area; this is the most critical step of
the procedure, since the only parameter for this decision is the thickness
of the rectal lesion. The cine-scan is generally targeted at the thickest
section of the tumor, although the highest blood flow in the tumoral
bed is expected to be peripheral, while poor blood supply, hypoxia, and
apoptosis more frequently occur at the center of the tumor. The areas with
9781842143094-Ch08 8/8/07 12:02 PM Page 131
Tumors of the gastro-intestinal tract 131
the highest blood supply cannot be chosen prior to performing CTP, and
sampling is one of the main factors differentiating CTP from the patho-
logical evaluation of tumor vascularization, since the pathologist selects
the area with the highest vascularization to evaluate MVD.
First-pass imaging requires high-frequency sequences (at least one scan/s)
up to 50–60 s after initiation of contrast medium injection, when the
contrast medium is predominantly intravascular. Reliable assessment of
microvessel blood volume (BV) and accurate determination of functional
parameters such as blood flow (BF) and mean transit time (MTT) can be
obtained from the first-pass imaging. Delayed imaging of rectal cancer, to
evaluate contrast medium leakage into extravascular space, requires scan-
ning up to 2–10 min, at a lower frequency than first-pass imaging; these
series allow microvessel permeability to be computed with reliable and
reproducible accuracy.19–21 The correct scanning protocol, including
both first-pass and delayed imaging, may be designed as follows: 8–10 s
scanning delay from the start of contrast medium intravenous injection
(contrast medium is still in the pulmonary circuit, and the first 10 seconds
are not relevant for tumor perfusion), followed by continuous scanning
at high time resolution (1 s/scan) for 45–50 s and at low time resolu-
tion (10–15 s/scan) up to 3–4 minutes. For dose reduction, 100 kVp and
200–220 mAs should be used; however, for oncological applications the
dose might not be relevant, since many rectal cancer patients undergo
radiation therapy.
High spatial resolution with thin slices is not required for perfusion
imaging; on the contrary, thick slices (5–10 mm) are preferred since they
reduce image noise, which affects perfusion measurements. The acqui-
sition protocol for a 16-slice scanner should include two adjacent 10-mm
slices or four 5-mm slices. Spiral acquisitions increase volume coverage,
but will produce unacceptable time resolution for the perfusion imaging
of rectal cancer.
The image acquisition protocol is summarized in Table 8.1.
Contrast medium administration

The great advantage of CTP imaging is the use of conventional contrast
medium, the same as in routine clinical activity, with concentrations
ranging from 300 to 370 mg/ml; the higher is the concentration, the
better is the tissue enhancement achieved.21
Similarly, the higher is the injection rate, the better is the tissue
enhancement and, hence, are the perfusion measurements. A bolus of
contrast medium ranging from 40 to 70 ml21 at high flow rate (4–7 ml/s)
is required for accurate first-pass measurements. Contrast medium is
administered via an 18–20-gauge cannula in the antecubital vein.
9781842143094-Ch08 8/8/07 12:02 PM Page 132
Table 8.1 Aquisition protocol
Slice thickness 4 × 5 mm or 2 × 10 mm
kVp 100
mAs 200–220
Dynamic scans Delay: 8–10 s
First pass: 45–50 s (one scan/s)
Delayed phase: 120–180 s (1 scan/10–15 s)
Contrast medium 300–370 mg/ml
40–70 ml
4-7 ml/s
18–20-gauge in the antecubital vein
DATA AND IMAGE ANALYSIS
The rationale for CT perfusion imaging is the repeated CT scanning of the

same volume, in order to obtain CT attenuation (HU, Hounsfield units)
curves over time (Figure 8.1). Several kinetic models have been applied
a 219
5
200
180
160
Artery
Hounsfield units
140
120
100
5
80 5
60 2
2 2
40 2 Tumor tissue
21
0 10000 20000 30000 40000 49001 ms
Time (ms)
Figure 8.1 Graphs representing time (seconds) on the x-axis and computed
tomography (CT) attenuation–density (HU) on the y-axis. Time–density curves
can be obtained for different locations in the CT image: artery (a),
9781842143094-Ch08 8/8/07 12:02 PM Page 133
b 62
60
55 Rectal cancer 3 (1313.0)
52
3 (1313.0)
48 3 (1313.0)
44
Hounsfield units
40 3 (1313.0)
36
32
28
24 5 (1313.0)
5 (1313.0)
20 5 (1313.0)
16 5 (1313.0)
12 5 (1313.0)
8 Rectal wall
5
0 10000 20000 30000 40000 49001 ms
Time (ms)
c 65
64 3
Muscle 3 3
62 3
60 2
58
4
56
4
54 4
Hounsfield units
2
52 Lymph node
50
48 2 2
46
44
Tumor tissue
42
40 2
38
36
0 10000 20000 30000 40000 49001 ms
Time (ms)
Figure 8.1 cont’d rectal cancer and normal rectal wall (b), muscle and lymph
node (c). They reflect CT attenuation over time.
9781842143094-Ch08 8/8/07 12:02 PM Page 134
by different authors to the time–density curves obtained to obtain per-

fusion measurements; compartmental analysis and deconvolution are
the most widely used for commercial application to body tumors.
Both have been largely validated22,23 on body tumors, and can be used
for rectal cancer CT perfusion imaging without any particular disadvan-
tage, since rectal anatomy is favorable, not influenced by respiratory or
other relevant motions (after hyoscin butylbromide administration).
Several dedicated software programs have been produced by different
vendors for data and image processing. The first step in perfusion
analysis is the accurate selection of arterial input. Small arteries will pro-
duce partial-volume effects; stenosis and circulation abnormalities will
also affect data. For correct arterial input placement in rectal cancer
perfusion imaging, the external iliac artery is recommended; the con-
tralateral artery may serve for partial-volume averaging.20,21 CT perfusion
software then generates a functional map for each computed perfusion
parameter, representing on a color scale the different values obtained
in the different parts of the CT image. A region of interest (ROI) can be
now placed on the rectal cancer and in other regions of the CT image,
to compute the perfusion values relative to the ROI area; manually
drawn ROIs are recommended for rectal cancer in order to accurately
include tumor margins, which are expected to have elevated angiogenesis.
Other ROIs can be placed on the normal rectal wall, for comparison
with rectal cancer values, and in adjacent organs (Figure 8.2).
ROIs placed on gluteus muscles, for example, may serve as controls
in perfusion measurements during therapy; changes of rectal cancer
perfusion parameters, expected to vary significantly after therapy,17 may
be compared to changes of muscle perfusion parameters, which are
expected to have constant perfusion values during therapy.
ROIs placed in adjacent organs, such as the prostate, lymph nodes,
bones, etc., may assess their involvement, with additional information
for rectal cancer imaging.
CURRENT EXPERIENCE AND CLINICAL APPLICATION
CT perfusion has already been used for rectal cancer assessment,16–18,24

with encouraging performances for diagnosis, risk stratification, and
therapy monitoring.
Tumor angiogenesis, which is the growth of new vessels from pre-
existing ones, results in an increased vascular bed, which may be reflected
in increased values of blood volume measured by CTP. Angiogenesis
also stimulates the opening of arteriovenous shunts, which have very
9781842143094-Ch08 8/8/07 12:02 PM Page 135
a b
c d
Figure 8.2 A 63-year-old patient with T3N1 rectal cancer. Regions of interest
(ROIs) are manually drawn on the original CT image (a): on rectal cancer
(lesion), normal rectal wall (normal wall), left external iliac artery (artery),
mesorectal pathologic lymph node (lymph node), and right gluteus muscle
(muscle). Functional maps of blood flow (b), blood volume (c), mean transit
time (d) and permeability surface area (e) are generated.
9781842143094-Ch08 8/8/07 12:02 PM Page 136
low resistance to flow and may result in an increased blood flow at

CTP, with a reduction of the mean transit time. Contrast medium
extravasation into extravascular space is expected to be high in rectal
cancer, due to the hyperpermeable nature of capillary endothelial
membranes of newly developed tumor vessels.
Significant differences have been reported between the perfusion
parameters of rectal cancer and those of the normal rectal wall. A study
of 15 patients17 showed significantly higher blood flow and lower mean
transit time in non-mucinous rectal adenocarcinoma, compared with
the values in the normal rectal wall; in the authors’ experience with a
25-patient cohort, BF, BV, and PS were significantly higher in rectal
cancer, compared to the normal rectal wall.25
Patients with rectal cancer undergoing neoadjuvant chemoradiation
therapy have different responses to treatment,26 but currently there are
no clinically useful predictors of response based on standard patholog-
ical assessment and immunocytochemistry.27 In the authors’ experience,
with the same 25-patient cohort blood flow and blood volume of rectal
cancers before neoadjuvant chemoradiation therapy were significantly
lower in non-responders than in responding patients ( p = 0.03 and
p = 0.003, respectively); rectal cancers with low blood flow and blood
volume may have less effective chemotherapeutic drug delivery and
lower oxygenation, with a related lower radiosensitivity, and thus a poor
response, as compared to those with high blood flow and blood volume.
Other authors17 have related high blood flow and short mean transit time,
which may reflect high intrinsic angiogenic activity in the tumor or a
secondary response to tissue hypoxia,28,29 to poor response.
However, CTP of rectal cancer shows potential for the prediction of
response to therapy; if further studies confirm these preliminary results,
CTP will play an important role in rectal cancer risk stratification, for
the best treatment selection.
CTP of rectal cancer may also play a role in therapy monitoring, with
a major impact on overall patient management, since it may solve
problems related to morphologic imaging (Figures 8.3 and 8.4). The
assessment of rectal cancer response to chemo- or radiation therapy with
morphologic imaging is mainly based on size criteria, with a time lag
between changes in tumor physiology induced by therapy, i.e. tumor
metabolism and perfusion, and changes in tumor morphology detectable
by imaging, which occur later. CTP detected significant changes in
rectal cancer perfusion after neoadjuvant chemoradiation therapy,17 and
is expected to detect early changes in rectal cancer perfusion during
therapy. This may potentially influence the course and duration of ther-
apy, which can be adjusted while treatment is still being administered,
9781842143094-Ch08 8/8/07 12:02 PM Page 137
a b
c
Figure 8.3 A 56-year-old woman
with T3N1 rectal cancer, before
neoadjuvant chemoradiation ther-
apy. Original CT image (a) and
functional maps of blood flow (b)
and blood volume (c).
customizing it to patient response. In addition, it might be possible to

predict the outcome of therapy at an early stage, and potentially avoid
unproductive and expensive treatments for some patients, or switch
them to a different therapy.
CTP of rectal cancer may play an even bigger role in the monitoring
of novel antiangiogenic therapies, which are increasingly being
adopted by different institutions for colorectal cancer treatment.4–6
Newly developed tumor vessels in the process of angiogenesis are too
small to be resolved using current diagnostic imaging techniques,
which are not reliable for monitoring changes in tumor vasculature fol-
lowing anti-angiogenic treatment. MVD assessment for such purpose is
limited for serial therapy monitoring in human patients, since it is an
invasive technique, requiring tissue sampling: furthermore, MVD counts
may not reflect the efficacy of the treatment, because it does not
independently measure vascular inhibition, but rather reflect many
different changes occurring in tumor vasculature over time.29 On the
contrary, changes of tumor microvasculature occurring in the process
9781842143094-Ch08 8/8/07 12:02 PM Page 138
a b
c
Figure 8.4 The same woman as
Figure 8.3, after neoadjuvant
chemoradiation therapy. Original CT
image demonstrates tumor size reduc-
tion (a). Functional maps did not
show any significant difference in
blood flow (b) and blood volume (c)
between tumor and normal rectal
wall: pathologic stage revealed
absence of tumor (T0), after neo-
adjuvant treatment.
of angiogenesis can be reliably assessed by CTP, with a reported pre-

cision ranging from 14% to 24%, for BF, BV, MTT and PS.19 Therefore,
CTP is suitable for monitoring changes occurring in tumor neovascu-
lature, following anti-angiogenic therapy; particularly, it is extremely
useful to detect early physiological changes occurring after therapy,
avoiding the time-lag of conventional imaging, which produces only
morphologic assessment of tumor volume changes following therapy,
which will occur later.31,32 In a study by Willett et al.,33 CTP has been
used in conjunction with other surrogate markers of angiogenesis,
such as MVD, interstitial fluid pressure (IFP), 18-fluorodeoxyglucose
(FDG) uptake, and systemic response (VEGF level in blood, number
of circulating endothelial cells (CECs) and progenitor cells), for the
assessment of changes in rectal cancer physiology induced by a single
infusion of the VEGF-specific antibody bevacizumab. CTP showed sig-
nificant changes after therapy, with results comparable to those using
9781842143094-Ch08 8/8/07 12:02 PM Page 139
MVD, IFP, and assessment of the number of CECs and progenitor

cells/stem cells.
There is, therefore, great interest in the integration of CTP into trials
of antiangiogenic therapy for rectal cancer. Furthermore, CTP presents
additional advantages, compared to other functional imaging tech-
niques. Radionuclide techniques are limited by poor spatial resolution
and shortcomings in the assessment of anatomic details; positron emis-
sion tomography (PET) remains an expensive imaging modality
because of the use of short-lived isotopes that require cyclotron pro-
duction and radiochemistry facilities on-site.34 Magnetic resonance (MR)
perfusion protocols may not be so easily understood by radiographers,
radiologists, clinicians, and research scientists as those of CTP; further-
more, MR is more expensive and not so widely available as CT.
Part B: Other gastrointestinal tumors

Kenneth A Miles
Cancer of the pancreas

CT perfusion is well suited to the measurement of pancreatic perfusion.
The fixed retroperitoneal position of the pancreas reduces the likeli-
hood of motion artefacts and the high spatial resolution means that the
pancreas can be reliably separated from the many adjacent vascular
structures that can adversely affect other perfusion imaging techniques
such as O-15 water positron emission tomography (PET).35 The first CT
perfusion study of the pancreas was reported in 1995,36 describing a
range of normal perfusion values between 125 and 166 ml/min/100 ml
along with findings of reduced perfusion in diabetes and increased
perfusion in an islet cell tumor (Figure 8.5). A later study by Tsushima
et al37 reported reductions in normal pancreatic perfusion with increas-
ing age with values as low as 55.4 ml/min/100 ml.
Abe et al have reported CT perfusion values from 8 pancreatic adeno-
carcinomas and one gastrin secreting islet cell tumor.38 Adenocarcinomas
demonstrated perfusion values lower than normal pancreas (median:
34.7 ml/min/100ml; range 22.1–50.0 ml/min/100 ml) whereas perfusion
values for the islet cell tumor were greater than normal pancreas
(196.2 ml/min/100ml) in keeping with the case previously reported by
Miles et al.36 Perfusion values also correlated closely with those derived
9781842143094-Ch08 8/8/07 12:02 PM Page 140
Figure 8.5 Conventional CT (a) and perfusion image (b) of a pancreatic islet
cell tumor. Note that perfusion in the tumor periphery is greater than in the
adjacent normal pancreas but perfusion is reduced in the center of the tumor.
(Reproduced with permission from reference 37)
9781842143094-Ch08 8/8/07 12:02 PM Page 141
using xenon CT39. Although using qualitative analysis only, Zhongqiu

et al. correlated the degree enhancement of pancreatic tumors relative
to surrounding normal tissue against histopathological features.40 Lower
relative enhancement was found in moderately or poorly differentiated
tumors, reflecting lower levels of perfusion.
The role of CT perfusion in pancreatic imaging remains to be estab-
lished. Potential applications include the monitoring of anti-cancer
therapy and the identification of sub-populations of patients in whom
anti-cancer therapy is more likely to be effective.39
Primary liver tumors

CT perfusion can be used to evaluate changes in hepatic perfusion
associated with hepatic cirrhosis.41 From time to time, such studies will
also depict hepatocellular carcinomas, reflecting the increased inci-
dence of this tumor in chronic liver disease (Figure 8.6). However, there
have been no systematic studies of CT perfusion in primary liver tumors.
Based on small series,37,42 primary liver tumors tend to exhibit perfusion
values greater than normal hepatic arterial perfusion. The mean perfu-
sion of two hepatocellular carcinomas was 65 ml/min/100 ml, consid-
erably lower than the mean perfusion from eight cases of focal nodular
hyperplasia at 136 ml/min/100 ml, highlighting a potential role for
CT perfusion in lesion characterization.
Figure 8.6 Conventional CT (a),

9781842143094-Ch08 8/8/07 12:02 PM Page 142
Figure 8.6 cont’d hepatic arterial (b), and hepatic portal (c) perfusion images
in a patient with hepatic cirrhosis and a hepatocellular carcinoma in the right
lobe. The tumor demonstrates high arterial perfusion. Portal perfusion is
reduced in the remaining liver reflecting portal hypertension. (Reproduced with
permission from reference 37)
REFERENCES
1. Birgisson H, Talback M, Gunnarsson U et al. Improved survival in cancer of

the colon and rectum in Sweden. Eur J Surg Oncol 2005; 31: 845–53.
2. American Cancer Society. Cancer reference information. 2005. Available at:
http://www.cancer.org/docroot/CRI/CRI_0.asp
9781842143094-Ch08 8/8/07 12:02 PM Page 143
3. Tepper JE, O’Connell M, Niedzwiecki D et al. Adjuvant therapy in rectal

cancer: analysis of stage, sex, and local control – final report of intergroup
0114. J Clin Oncol 2002; 20: 1744–50.
4. O’Connell MJ. Current status of adjuvant therapy for colorectal cancer.
Oncology 2004; 18: 751–5.
5. Zhu AX, Willett CG. Combined modality treatment for rectal cancer. Semin
Oncol 2005; 32: 103–12.
6. Hobday TJ. An overview of approaches to adjuvant therapy for colorectal
cancer in the United States. Clin Colorectal Cancer 2005; 5: S11–18.
7. Takebayashi Y, Aklyama S, Yamada K et al. Angiogenesis as an unfavorable
prognostic factor in human colorectal carcinoma. Cancer 1996; 78: 226–31.
8. Li C, Gardy R, Seon BK et al. Both high intratumoral microvessel density
determined using CD105 antibody and elevated plasma levels of CD105 in
colorectal cancer patients correlate with poor prognosis. Br J Cancer 2003; 88:
1424–31.
9. Cascinu S, Graziano F, Catalano V et al. An analysis of p53, BAX and vascular
endothelial growth factor expression in node-positive rectal cancer.
Relationships with tumour recurrence and event-free survival of patients
treated with adjuvant chemoradiation. Br J Cancer 2002; 86: 744–9.
10. Theodoropoulos GE, Lazaris AC, Theodoropoulos VE et al. Hypoxia, angio-
genesis and apoptosis markers in locally advanced rectal cancer. Int J
Colorectal Dis 2006; 21: 248–57.
11. Li WW. Tumor angiogenesis: molecular pathology, therapeutic targeting, and
imaging. Acad Radiol 2000; 7: 800–11.
13. Tateishi U, Nishihara H, Watanabe S et al. Tumor angiogenesis and dynamic
CT in lung adenocarcinoma: radiologic-pathologic correlation. J Comput Assist
Tomogr 2001; 25: 23–7.
14. Tateishi U, Kusumoto M, Nishihara H et al. Contrast-enhanced dynamic com-
puted tomography for the evaluation of tumor angiogenesis in patients with
lung carcinoma. Cancer 2002; 95: 835–42.
15. Miles KA. Tumour angiogenesis and its relation to contrast enhancement on
computed tomography: a review. Eur J Radiol 1999; 30: 198–205.
16. Harvey C, Dooher A, Morgan J et al. Imaging of tumour therapy responses by
dynamic CT. Eur J Radiol 1999; 30: 221–6.
Radiology 2005; 234: 785–92.
18. Goh V, Halligan S, Hugill JA, Gartner L, Bartram CI. Quantitative colorectal
cancer perfusion measurement using dynamic contrast-enhanced multidetector-
row computed tomography: effect of acquisition time and implications for
protocols. J Comput Assist Tomog 2005; 29: 59–63.
19. Purdie TG, Henderson E, Lee TY. Functional CT imaging of angiogenesis in
rabbit VX2 soft tissue. Phys Med Biol 2001; 46: 3161–75.
20. Lee TY, Purdie TG, Stewart E. CT imaging of angiogenesis. Q J Nucl Med 2003;
47: 171–187.
21. Miles KA. Perfusion CT for the assessment of tumour vascularity: which
protocol? Br J Radiol 2003; 76: S36–42.
9781842143094-Ch08 8/8/07 12:02 PM Page 144
obtained with dynamic CT. Radiology 1993; 188: 405–11.
23. Nabavi DG, Cenic A, Dool J et al. Quantitative assessment of cerebral hemo-
dynamics using CT: stability, accuracy, and precision studies in dogs. J Comput
Assist Tomogr 1999; 23: 506–15.
24. Harvey C, Morgan J, Blomley M et al. Tumor responses to radiation therapy:
use of dynamic contrast material-enhanced CT to monitor functional and
anatomical indices. Acad Radiol 2002; 9: S215–19.
25. Bellomi M, Petralia G, Sonzongni A, Zampino MG, Rocca A. CT perfusion for
the monitoring of neo-adjuvant chemoradiation therapy in rectal carcinoma –
initial experience. Radiology 2007; in press.
26. Rodel C, Martus P, Papadoupolos T et al. Prognostic significance of tumor
regression after preoperative chemoradiotherapy for rectal cancer. J Clin
Oncol 2005; 23: 8688–96.
27. Smith FM, Reynolds JV, Miller N et al. Pathological and molecular predictors
of the response of rectal cancer to neoadjuvant radiochemotherapy. Eur J Surg
Oncol 2006; 32: 55–64.
28. Leek RD, Landers RJ, Harris AL et al. Necrosis correlates with high vascular
density and focal macrophage infiltration in invasive carcinoma of the breast.
Br J Cancer 1999; 79: 991–5.
29. Wheeler RH, Ziessman HA, Medvec BR et al. Tumor blood flow and systemic
shunting in patients receiving intra-arterial chemotherapy for head and neck
cancer. Cancer Res 1986; 46: 4200–4.
30. Hlatky L, Hahnfeldt P, Folkman J. Clinical application of antiangiogenic therapy:
microvessel density, what it does and doesn’t tell us. J Natl Cancer Inst 2002;
94: 883–93.
31. Miles KA, Leggett DAC, Kelley BB et al. In-vivo assessment of neovascularisation
of liver metastases using perfusion CT. Br J Radiol 1998; 71: 276–81.
32. Mooteri S, Rubin D, Leurgans S et al. Tumor angiogenesis in primary and
metastatic colorectal cancers. Dis Colon Rectum 1996; 39: 1073–80.
Nat Med 2004; 10: 145–7.
34. Blankenberg FG, Eckelman WC, Strauss HW et al. Role of radionuclide imaging
in trials of antiangiogenic therapy. Acad Radiol 2000; 7: 851–67.
35. Bacharach SL, Libutti SK, Carrasquillo JA. Measuring tumor blood flow with
H(2)(15)O: practical considerations. Nucl Med Biol 2000; 27: 671–6.
36. Miles KA, Hayball MP, Dixon AK. Measurement of human pancreatic perfusion
using dynamic computed tomography with perfusion imaging. Br J Radiol
1995; 68: 471–5.
37. Miles K, Blomley M. Applications of perfusion CT. In: Miles K, Blomley M,
Dawson P, eds. Functional Computed Tomography. Oxford: ISIS Medical
Media, 1997: 89–116.
38. Tsushima Y, Kusano S. Age-dependent decline in parenchymal perfusion in
the normal human pancreas: measurement by computed tomography. Pancreas
1998; 17: 148–52.
39. Abe H, Murakami T, Kubota M et al. Quantitative tissue blood flow evaluation
of pancreatic tumor: comparison between xenon CT technique and
perfusion CT technique based on deconvolution analysis. Radiat Med 2005;
23: 364–370.
9781842143094-Ch08 8/8/07 12:02 PM Page 145
40. Zhongqiu W, Guangming L, Jieshou L et al. The comparative study of tumor

angiogenesis and CT enhancement in pancreatic carcinoma. Eur J Radiol 2004;
49: 274–80.
41. Van Beers BE, Leconte I, Materne R et al. Hepatic perfusion parameters in
chronic liver disease: dynamic CT measurements correlated with disease severity.
AJR Am J Roentgenol 2001; 176: 667–73.
42. Groell R, Kugler C, Aschauer M et al. Quantitative perfusion parameters of focal
nodular hyperplasia and normal liver parenchyma as determined by electron
beam tomography. 1995; 68: 1185–93.
9781842143094-Ch08 8/8/07 12:02 PM Page 146
9781842143094-Ch09 8/8/07 4:23 PM Page 147
9
Tumors of the urogenital tract
Elizabeth P Ives and Ethan J Halpern
For the 2007 calendar year, it is estimated that there will be 218 890 new
prostate cases and 27050 prostate cancer deaths in the United States.1 The
prostate is the leading site of new cancer diagnoses among US men and
the second leading cause of cancer deaths. For an American male, there
is a 1 in 6 lifetime probability of developing prostate cancer.
Traditional diagnostic imaging methods have a limited role in the
diagnosis, staging, and management of prostate cancer. The role of
imaging in the prostate is limited by poor discrimination of prostate
cancer from adjacent benign tissue, and by benign hypertrophic
changes within the prostate. Technological innovations in imaging have
the potential to improve the detection of prostate cancer with more
accurate biopsy techniques, better local staging, more precise focusing
of therapy, and closer post-treatment follow-up. Although computed
tomography (CT) currently has a small role in the diagnosis and stag-
ing of prostate disease,2 numerous reports indicate the frequent use
of CT by clinicians for evaluation of prostate cancer patients.3,4 Recent
improvements in multidetector CT technology permit quantitative
perfusion imaging of the prostate, and may presage a new role for CT
in the detection, staging, and management of prostate cancer.
Screening for prostate cancer is a controversial topic. Prostate spe-
cific antigen (PSA) serum levels and the digital rectal examination are
the current accepted standard of care for prostate cancer screening.
Transrectal ultrasound (TRUS) is not accepted as a screening method in
an unselected population because of the poor sensitivity and specificity
of sonography for prostate cancer. The accepted application for TRUS
is to guide a biopsy procedure of the prostate in a patient with an
abnormal serum PSA or digital rectal examination. Even if ultrasound-
guided biopsy can improve detection, it provides little information
about the extent of disease, because location and quantity of malig-
nancy on biopsy often do not correlate with actual cancer volume.5
9781842143094-Ch09 8/8/07 4:23 PM Page 148
The metabolic changes that allow neoplastic cells to develop into a

solid mass have been investigated extensively. Hypoxia from tumor
growth is thought to lead to angiogenesis factors which increase
microvascularity and allow the growth of solid tumors.6 Prostate cancer
demonstrates significantly higher microvessel density compared with
benign tissue.7 As with other organ systems, the microvessel density of
prostate cancer is useful in predicting the aggressiveness of the tumor.
It has been suggested that quantitative microvessel density can serve as
a prognostic marker for prostate cancer.8
Recent studies have exploited the alterations in perfusion in neoplastic
prostate tissue to develop imaging techniques for the detection, local-
ization, and management of prostate cancer. Perfusion imaging of the
prostate has been evaluated primarily with ultrasound and magnetic
resonance (MR) technologies. There is relatively little literature on the
application of CT perfusion imaging to the prostate.
Perfusion imaging with sonography is a key focus of research, since
ultrasound is used routinely to guide biopsy. Conventional Doppler
ultrasound has shown mixed results when using color and/or power
Doppler to target malignant tissue for biopsy.9 Microbubble contrast
agents have been proposed as a method to improve the detection of
prostate cancer.10 Gas encapsulated microbubbles remain within the vas-
culature for several minutes, and may be used as a contrast agent during
an ultrasound examination. Microbubbles show parenchymal organ
enhancement, and can be used to analyze the prostate blood flow pat-
terns.11 Several studies show this technique to increase the sensitivity of
core needle biopsy in detecting cancer.12–14 Studies using time–intensity
data from microbubble contrast sonography suggest that a quick time to
peak enhancement may be predictive of prostate malignancy.15
MR imaging (MRI) may improve the local staging of prostate cancer,16
and may be useful for targeted biopsy in patients with suspected cancer
and negative prior biopsy.17 In current clinical practice, however, MR
does not reliably alter staging and therapy based on traditional clinical
staging criteria. MR is primarily utilized to evaluate for disease exten-
sion beyond the prostate. Several studies have demonstrated that
prostate cancer tissue enhances earlier than normal tissue when using
a dynamic contrast-enhanced MRI sequence.18–20
The normal appearance of the prostate on CT is a smooth contoured
structure at the base of the pelvis bordered by the pubic symphysis,
bladder, and rectum. After the administration of intravenous iodinated
contrast, the transition zone of the prostate enhances brightly in the
inner gland (Figure 9.1). This enhancement can appear heterogeneous
with benign hyperplasia (Figure 9.2). The outer gland enhances to
9781842143094-Ch09 8/8/07 4:23 PM Page 149
Tumors of the urogenital tract 149
Figure 9.1 Contrast-enhanced computed tomography (CT) of the prostate

demonstrates enhancement of the transition zone in this normal prostate
(arrows)
a b
Figure 9.2 Patient with enlarged prostate secondary to benign prostatic hyper-
plasia. The pre-contrast image (a) demonstrates an enlarged gland. After con-
trast administration there is heterogeneous enhancement of the hypertrophied
transition zone that occupies most of the prostate (b)
9781842143094-Ch09 8/8/07 4:23 PM Page 150
a lesser degree and appears homogeneous in the normal prostate.

Outer gland tissue can often be distinguished clearly from inner gland
tissue based upon this enhancement pattern. Capsular enhancement
appears as a ring around the prostate gland.21
Prostate cancer is generally not visible with CT imaging. Prostate
cancer will not be detected on a CT scan performed without an intra-
venous contrast agent unless there is significant contour irregularity or
bulging (Figure 9.3). In a minority of patients it is possible to detect
prostate cancer based upon contrast enhancement (Figure 9.4). Based
upon recent advances in helical CT technology, it is possible to obtain
color-coded quantitative perfusion images corresponding to conventional
contrast-enhanced CT images (Figure 9.5).
The earliest published clinical study related to detection of prostate
cancer with CT perfusion used a single-detector helical CT with arterial
phase imaging to investigate 35 patients who had undergone TRUS
biopsy.22 Twenty-five of these patients were diagnosed with prostate
cancer based upon the biopsy. Ten patients were suspected of having
prostate cancer based on examination and PSA but had a negative biopsy.
Figure 9.3 Contour abnormality along the right side of the prostate corre-
sponds to an exophytic prostate cancer at the right mid-gland (arrow). Note that
this area of cancer does not enhance to a greater degree than other unaffected
parts of the prostate
9781842143094-Ch09 8/8/07 4:23 PM Page 151
a b
Figure 9.4 High-grade prostate cancer. Pre-contrast image (a) demonstrates a

normal homogeneous appearance of the prostate. After contrast administration
(b) there is focal enhancement of the cancer along the right side of the prostate
(arrow)
Figure 9.5 Normal CT and quantitative perfusion images. Images through the
mid-gland level of the prostate in a patient with a low-volume tumor in the apex
of the gland. Both the contrast-enhanced CT (a) and the quantitative perfusion
(b) demonstrate a normal, symmetric appearance
9781842143094-Ch09 8/8/07 4:23 PM Page 152
Areas of the prostate that demonstrated asymmetric focal or diffuse

contrast enhancement in the peripheral zone were considered positive
by CT. The authors report identifying 22 of 25 patients (88%) with proven
prostate cancer based upon increased arterial phase enhancement,
presumably related to increased perfusion. In the same 25 patients,
102 peripheral zone cancer sites were found on biopsy. CT identified
increased enhancement in 59/102 (58%) of the positive peripheral zone
sites. The study noted that patients with higher PSA levels (> 10 ng/ml)
and higher Gleason scores (> 5) had a greater degree of enhancement
and size of nodular changes than other patients. While the authors con-
cede that these results do not warrant routine use of CT for screening,
they did propose the use of CT in selected patient populations such as
those with rising PSA and negative biopsies, those with a rising PSA
following abdominal perineal resection, and patients with routine CT
pelvic examinations on which abnormal focal contrast enhancement in
the peripheral zone is observed. CT might be used in these populations
to guide a biopsy procedure of the prostate.
Based upon the reported sensitivity of CT perfusion for prostate
cancer using a single-detector helical scanner with qualitative assess-
ment of CT perfusion, the authors of this chapter performed a CT study
at Thomas Jefferson University with a 16-slice multidetector helical
scanner using quantitative CT perfusion.23 Ten patients with biopsy-
proven prostate cancer underwent contrast-enhanced pelvic CT prior
to radical prostatectomy. CT evaluation of the prostate was repeated at
10-second intervals during the first minute after administration of intra-
venous contrast. Six regions of interest (ROIs) were defined on the CT
images in a sextant distribution pattern (base, mid, and apex along the
right and left peripheral zones). Perfusion data were computed based
upon the maximum slope of enhancement in the prostate ROI relative
to the CT density of an arterial segment ROI.24 Perfusion data were cor-
related with whole-mount prostatectomy specimens, and were analyzed
for correlation with the presence of malignancy, Gleason score, and
percentage of tumor within the region on pathology. Notably, only one
of ten subjects demonstrated visible peripheral gland enhancement
(Figure 9.6). This patient demonstrated a Gleason 10 tumor with high
volume and a significant correlation between perfusion and pathology
results. Despite a lack of visible enhancement, one other subject had a
significant correlation between perfusion and pathology. This subject also
demonstrated aggressive (Gleason 8), high-volume (90% in at least one
region) disease (Figure 9.7). The remaining eight subjects with prostate
cancer showed neither visible enhancement nor significant correlation of
perfusion and pathology (Figure 9.8).
9781842143094-Ch09 8/8/07 4:23 PM Page 153
Figure 9.6 High-volume, Gleason 10 prostate cancer. The contrast-enhanced

image (a) demonstrates bilateral capsular enhancement, but more substantial
enhancement along the right mid-gland within the region of interest marked
T1, as compared to the left mid-gland. Quantitative perfusion (b) demonstrates
increased flow along the right mid-gland (arrow), corresponding to the site of
tumor
Based upon the study performed at Thomas Jefferson University, we

suggest that CT perfusion may have a role in the evaluation of high
volume, high grade prostate cancer, but does not have a role in the
detection or staging of most prostate cancer. Despite the disparate con-
clusions of the two CT studies reported in this chapter, both studies
suggest that high volume, aggressive prostate cancer does have
increased perfusion and may enhance on CT.
One group has published several reports related to the imaging of
prostate tumor response to radiation therapy on functional CT.25,26
These investigators evaluated subjects with biopsy-proven prostate cancer.
9781842143094-Ch09 8/8/07 4:23 PM Page 154
Figure 9.7 High-volume, Gleason 8 prostate cancer. The contrast-enhanced

image (a) demonstrates no apparent asymmetry in the enhancement pattern.
The quantitative perfusion image (b) demonstrates greater perfusion in the right
mid-gland, corresponding to the site of tumor (arrow)
Patients underwent functional CT prior to radiation therapy (RT) and

again at 1–2 weeks and 6–12 weeks post-treatment. ROIs drawn on the
entire prostate gland and on a large artery in the same location at each
time-point were used to produce time–attenuation curves. The prostate
perfusion, mean contrast agent clearance per unit volume, and the
mean fractional vascular volumes were calculated and compared over
time. The results demonstrated significant increases in all three param-
eters between the pre-RT scan and the 1–2-week post-therapy time-
point. No significant change in perfusion was shown between the
1–2-week time-point and the 6–12-week time-point.
9781842143094-Ch09 8/8/07 4:23 PM Page 155
Figure 9.8 Images of the prostate apex in a patient with a high-volume,

Gleason 6 tumor in the right apex. Contrast-enhanced CT (a) demonstrates
symmetric enhancement of the prostate, and quantitative perfusion (b) demon-
strates no difference in perfusion between the malignancy on the right and
benign tissue on the left side
Post-RT studies evaluated changes in CT perfusion of the prostate in

response to radiation treatment. The results demonstrated early hyper-
emia of the gland, a known response of tissue to radiation. Since
prostate cancer may also appear hyperemic, the authors could not deter-
mine whether prostate cancer perfusion responded differently to radia-
tion therapy as compared with normal tissue. PSA values would still
be necessary to follow disease response, and cannot be replaced by
CT perfusion. The hyperemic response was still evident at 6–12 weeks,
and gave no indication of the overall effect of RT on the disease process.
Further studies are needed to clarify the clinical utility of CT perfusion
after therapy for prostate cancer.
Based upon currently available clinical studies, there is no clear-cut
clinical indication for CT perfusion in the evaluation of prostate cancer.
Unlike many solid tumors which tend to grow as a solitary round mass,
9781842143094-Ch09 8/8/07 4:23 PM Page 156
prostate cancer is commonly multifocal, is often infiltrative throughout

the prostate, and may grow in an oblong or irregular shape along the
capsule of the prostate gland.27,28 These histologic features of prostate
cancer growth limit our ability to visualize prostate cancer with conven-
tional imaging of the prostate by ultrasound, CT, and MRI. These same
histologic features are likely to complicate the results of CT perfusion.
Future research is required to determine whether a more sensitive
perfusion technique may improve the detection and staging of prostate
cancer.
PERFUSION IMAGING OF OTHER TUMORS OF THE

UROGENITAL TRACT
Kenneth A Miles
Renal cancer
CT perfusion studies of renal cancer were first reported in 1994.29 Renal
tumors are typically highly vascularized, but CT perfusion images
frequently show high perfusion peripherally with lower perfusion in a
central area of cystic change or necrosis29,30 (Figure 9.9). Metastatic
lesions from renal cancer are also frequently hypervascular (Figure 9.10).
Figure 9.9 CT perfusion image of renal cancer. (Reproduced with permission

from reference 31)
9781842143094-Ch09 8/8/07 4:23 PM Page 157
Figure 9.10 Hepatic arterial (a) and portal (b) perfusion images of a hepatic
metastasis from renal cancer. Arterial perfusion values are very high, whereas
there is virtually no portal perfusion in the lesion. Note also abnormal arterial
and portal perfusion values in the adjacent liver tissue that is apparently normal
on conventional CT images. (Reproduced with permission from reference 31)
Due to the mobility of the kidneys, particular care must be taken to

control respiratory motion when generating CT perfusion images of
renal tumors.
The recent dynamic contrast-enhanced CT study of 24 renal cancers
by Wang et al. reports a range of enhancement parameters which
demonstrate significant correlations with microvessel density.30 The
rapid cycle time (one image every 4.9 s) allows further analysis of the
9781842143094-Ch09 8/8/07 4:23 PM Page 158
12
10
SPV 8
0
50 100 150 200
MVD (/mm2)
Figure 9.11 Correlation between standardized perfusion value (SPV) and

microvessel density (MVD) based on the results of Wang et al.30 r = 0.48, p < 0.02
reported enhancement parameters to derive estimated perfusion values

(slope method) and standardized perfusion values (SPVs).32 Using an
iodine calibration factor of 21.32 HU/mg iodine,33 the medians (ranges)
for tumor perfusion and SPV were 285 (47–1140) ml/min/100 ml and
2.0 (0.2–11) ml/min/100 ml respectively. The SPV was also found to
correlate significantly with microvessel density (r = 0.48, p < 0.02;
Figure 9.11), whereas no significant correlation was seen with estimated
perfusion (r = 0.23, p = 0.28). Because the SPV reflects tumor perfusion
normalized to cardiac output,32 SPV values are more likely to reflect
microvessel density than are perfusion values, which are affected by
central circulatory parameters.
The earlier study of 40 renal tumours by Jinzaki et al. also found a
significant correlation between tumor enhancement at 35 s and microves-
sel density.34 These researchers additionally reported significantly higher
enhancement in clear-cell tumors as compared to other types of renal
cancer (Table 9.1). Clear-cell tumors all showed enhancement of greater
than 100 HU (3.3 HU/g iodine), whereas the remaining tumors enhanced
below this threshold.
Cervical cancer
There has been a single but thorough study of CT perfusion in cancer
of the uterine cervix.35 The study group comprised 32 patients with
tumor stages ranging between T1b and T3b. Tumors demonstrated mod-
erately high vascularization. Based on deconvolution analysis, mean values
for perfusion, blood volume, and permeability were 95.5 ml/min/100 g,
9781842143094-Ch09 8/8/07 4:23 PM Page 159
Table 9.1 Peak enhancement values (35 s or 180 s) and corresponding perfusion/
cardiac output for various types of renal cancer (RCC) based on reference 34
Tumor type Peak enhancement (HU) Perfusion/cardiac output (%/100 ml)*
Clear-cell RCC 165.0 ± 45.8 2.58 ± 0.72

Chromophobe RCC 91.6 ± 1.6 1.43 ± 0.03
Papillary RCC 62.5 ± 7.0 0.98 ± 0.11
Oncocytoma 94.2 ± 22.3 1.47 ± 0.35
Metanephric adenoma 92.5 ± 7.0 1.45 ± 0.11
*Assumes an iodine calibration factor of 21.32 HU/mg iodine as reported by Miles et al.33
21.6 ml/100 g, and 18.0 ml/min/100 g respectively. Blood volume and

permeability values were significantly higher in patients with enlarged
locoregional lymph nodes on magnetic resonance imaging, suggesting
a link between intensity of angiogenesis and propensity for tumor
metastasis. Perfusion values demonstrated a moderate positive correla-
tion with measurements of tumor oxygenation. Tumor hypoxia is
known to be an adverse prognostic factor for cervical cancer. Although
further validation is required, this study has highlighted the potential for
CT perfusion to act as a less invasive prognostic marker than the
polarographic needle electrode system required to measure tumor
oxygen tension.
REFERENCES
1. Jemal A, Siegel R, Ward E et al. Cancer statistics, 2007. CA Cancer J Clin 2007;
57: 43–66.
2. Levran A, Gonzalez, JA, Diokno AC et al. Are pelvic computed tomography,
bone scan, and pelvic lymphadenectomy necessary in the staging of prostatic
cancer? Br J Urol 1995; 75: 778–81.
3. Kindrick AV, Grossfeld GD, Stier DM et al. Use of imaging tests for staging
newly diagnosed prostate cancer: trends from the CAPSURE database. Urology
1998; 160: 2102–6.
4. Saigal CS, Pashos CL, Henning JM, Litwin MS. Variations in use of imaging in a
national sample of men with early-stage prostate cancer. Urology 2002; 59: 400–4.
5. Gardner TA, Lemer ML, Schlegel RS et al. Microfocal prostate cancer: biopsy
cancer volume does not predict actual tumour volume. Br J Urol 1998; 81:
839–43.
6. Dang CV, Semenza GL. Oncogenic alterations of metabolism. Trends Biochem
Sci 1999; 24: 68–72.
7. Bigler SA, Deering RE, Brawer MK. Comparison of microscopic vascularity in
benign and malignant prostate tissue. Human Pathol 1993; 24: 220–6.
9781842143094-Ch09 8/8/07 4:23 PM Page 160
8. Brawer MK. Quantitative microvessel density: a staging and prognostic marker

for human prostatic carcinoma. Cancer 1996; 78: 345–9.
9. Halpern EJ, Strup SE. Using gray scale, color and power Doppler sonography
to detect prostatic cancer. AJR Am J Roentgenol 2000; 174: 623–7.
10. Wijkstra H, Wink MH, de la Rosette JJ. Contrast specific imaging in the detec-
tion and localization of prostate cancer. World J Urol 2004; 22: 346–50.
11. Aarnink RG, Beerlage HP, de la Rosette JJ et al. Contrast angiosonography:
a technology to improve Doppler ultrasound examinations of the prostate.
Eur Urol 1999; 35: 9–20.
12. Halpern EJ, Verkh L, Forsberg F et al. Initial experience with contrast-enhanced
sonography of the prostate. AJR Am J Roentgenol 2000; 174: 1575–80.
13. Halpern EJ, Rosenberg M, Gomella LG. Prostate cancer: contrast enhanced US
for detection. Radiology 2001; 219: 219–25.
14. Halpern EJ, Ramey JR, Strup SE et al. Detection of prostate cancer with
contrast enhanced sonography using intermittent harmonic imaging. Cancer
2005; 104: 2372–83.
15. Goossen TE, de la Rosette JJ, Hulsbergen-van de Kaa CA, van Leenders GJ,
Wijkstra H. The value of dynamic contrast enhanced power Doppler ultrasound
imaging in the localization of prostate cancer. Eur Urol 2003; 43: 124–31.
16. Claus FG, Hricak H, Hattery RR. Pretreatment evaluation of prostate cancer: role
of MR imaging and 1H MR spectroscopy. Radiographics 2004; 24 (Suppl 1):
S167–80.
17. Prando A, Kurhanewicz J, Borges AP, Oliveira EM Jr, Figueiredo E. Prostatic
biopsy directed with endorectal MR spectroscopic imaging findings in patients
with elevated prostate specific antigen levels and prior negative biopsy findings:
early experience. Radiology 2005; 236: 903–10.
18. Brown G, Macvicar DA, Ayton V et al. The role of intravenous contrast
enhancement in magnetic resonance imaging of prostatic carcinoma. Clin Radiol
1995; 50: 601–6.
19. Jager GJ, Ruijter E, van de Kaa CA et al. Dynamic TurboFLASH subtraction
technique for contrast-enhanced MR imaging of the prostate: correlation with
histopathologic results. Radiology 1997; 203: 645–51.
20. Engelbrecht MR, Huisman HJ, Laheij R et al. Discrimination of prostate cancer
from normal peripheral zone and central gland tissue by using dynamic
contrast-enhanced MR imaging. Radiology 2003; 229: 248–54.
21. Halpern EJ, Cochlin DL, Goldberg BB. Imaging of the Prostate. New York:
Martin Dunitz, 2002.
22. Prando A, Wallace S. Helical CT of prostate cancer: early clinical experience.
AJR Am J Roentgenol 2000; 175: 343–6.
23. Ives EP, Burke MA, Edmonds PR et al. Quantitative CT perfusion of prostate
cancer: correlation with whole mount pathology. Clin Prostate Cancer 2005;
4: 109–12.
Br J Radiol 1991; 64: 409–12.
25. Harvey C, Dooher A, Morgan J et al. Imaging of tumour therapy responses by
26. Harvey CJ, Blomley MJ, Dawson P et al. Functional CT imaging of the acute
hyperemic response to radiation therapy of the prostate gland: early experience.
9781842143094-Ch09 8/8/07 4:23 PM Page 161
27. Byar DP, Mostofi FK. Carcinoma of the prostate: prognostic evaluation of certain
pathologic features in 208 radical prostatectomies. Cancer 1972; 30: 5–13.
28. McNeal JE, Redwine EA, Freiha FS, Stamey TRA. Zonal distribution of prostatic
adenocarcinoma. Correlation with histologic pattern and direction of spread.
Am J Surg Pathol 1988; 12: 897–906.
29. Miles KA, Hayball MP, Dixon AK. Functional imaging of changes in human
intra-renal perfusion using quantitative dynamic computed tomography.
Invest Radiol 1994; 29: 911–14.
30. Wang JH, Min PQ, Wang PJ et al. Dynamic CT evaluation of tumor vascularity
in renal cell carcinoma. AJR Am J Roentgenol 2006; 186: 1423–30.
Media, 1997: 89–116.
Radiology 2001; 220: 548–53.
919–26.
34. Jinzaki M, Tanimoto A, Mukai M et al. Double-phase helical CT of small renal
parenchymal neoplasms: correlation with pathologic findings and tumor
angiogenesis. J Comput Assist Tomogr 2000; 24: 835–42.
35. Haider MA, Milosevic M, Fyles A et al. Assessment of the tumor microenviron-
ment in cervix cancer using dynamic contrast enhanced CT, interstitial fluid
pressure and oxygen measurements. Int J Radiat Oncol Biol Phys 2005; 62:
1100–7.
9781842143094-Ch09 8/8/07 4:23 PM Page 162
9781842143094-Ch10 8/8/07 12:05 PM Page 163
10
CT perfusion of lymph nodes
Kenneth A Miles
INTRODUCTION
The assessment of lymph nodes is an essential component of the appli-

cation of multidetector computed tomography (CT) in oncology, both
in the detection of metastatic disease within the locoregional lymph
nodes draining a tumor, and in the assessment of lymphoma. The main
criterion used in conventional CT assessments comprises a measure-
ment of nodal size, although other morphological features such as
shape and texture, and the appearances of nearby nodes, may also
contribute. However, the limitations of this morphological approach are
well recognized. For example, lymph nodes that contain tumor may be
smaller than the cut-off size used for the diagnosis of involvement.
On the other hand, lymph nodes may be enlarged yet be reactive rather
than contain tumor cells. Lymph nodes may also remain enlarged despite
successful treatment (see Figure 10.5). Furthermore, interobserver varia-
tion in the measurement of lymph node size creates significant potential
for misclassification.1
In view of these limitations, a range of functional imaging techniques
has been explored as a means to improve the accuracy of diagnosis of
lymph node involvement by tumor, most notably the use of fluoro-
deoxyglucose-positron emission tomography (FDG-PET) using integrated
PET–CT systems. However, the functional information obtainable
with CT perfusion has the potential to provide similar benefits but
with greater simplicity and availability than with PET-CT. When
located within body regions that are less susceptible to respiratory
motion, for example the retroperitoneum and lung, lymph nodes are
frequently well suited for reliable measurement of enhancement or per-
fusion (Box 10.1). However, measurements may be difficult in small
lymph nodes (i.e. ≤ 5 mm). Early in the first pass of contrast medium,
9781842143094-Ch10 8/8/07 12:05 PM Page 164
Box 10.1 Perfusion CT of lymph nodes: technical issues
● less prone to respiratory motion e.g. mediastinum, retroperitoneum

● measurements may be difficult in small nodes (< 5 mm)
● potential artifacts when adjacent to large vessels
● assessment of lymph nodes or primary tumor.
high vascular contrast density can also produce beam hardening artifacts
within lymph nodes that lie close to major vessels. Although the appli-
cation of CT perfusion to the assessment of tumor within lymph nodes
is, as yet, underdeveloped, this chapter reviews the currently available
data supporting the use of this technique, both in the diagnosis of
lymph node metastases and in the assessment of lymphoma.
LYMPH NODE METASTASES
Three studies have shown the potential for CT perfusion to improve

the performance of CT in the diagnosis of lymph node metastases
(Table 10.1). All three studies had pathological confirmation of nodal
status and used measures of lymph node enhancement which could be
shown to reflect perfusion relative to cardiac output (see Chapter 3).
The first of these studies considered the staging of 58 patients with
gastric cancer.2 A helical data acquisition was commenced at 40 seconds
following a 100-ml bolus of contrast material with an iodine concentra-
tion of 300 mg/ml injected at 3 ml/s. The tube voltage for image acqui-
sition was 120 kVp. When a nodal attenuation of 100 HU was used as
Table 10.1 Comparison of the diagnostic performances of size (single node ≥ 10 mm),
enhancement, and combined criteria in the diagnosis of nodal metastases
Enhancement Combined
Size criteria criteria criteria
Primary Sens. Spec. Sens. Spec. Sens. Spec.
tumor Authors (%) (%) (%) (%) (%) (%)
Stomach Fukuya et al.2 33 78 64 93 85 88

Breast Yuen et al.3 15 100 78 52 89 93
Lung (T1) Shim et al.4 27 90 65* 76* 88* 90*
*Enhancement of primary tumor; Sens., sensitivity; Spec., specificity

9781842143094-Ch10 8/8/07 12:05 PM Page 165
CT perfusion of lymph nodes 165
the threshold to distinguish metastasis-positive and metastasis-negative

nodes, the sensitivity was 64% and specificity 93%. This diagnostic
performance was superior to that obtained using size criteria, but was
further improved when size and enhancement criteria were used in
combination. As no pre-contrast images were obtained, true enhance-
ment measures were not available. However, assuming a mean baseline
nodal attenuation of 25 HU as observed by Yuen et al.,3 this enhance-
ment threshold can be approximated to 75 HU or 2.5 HU/g iodine
injected.
Yuen et al. studied the enhancement of the sentinel axillary node in
107 Japanese women with breast cancer ranging from cancer-in-situ to
stage T4.3 Images were acquired pre-contrast and at 1, 3, and 8 minutes
after a 27-g load of iodinated contrast material injected over 45 s.
A higher tube voltage of 135 kVp was used, and thus a lesser degree of
enhancement would be expected for a given iodine load compared to
120 kVp (Chapter 3). There were significant differences in the average
enhancement curves observed in metastasis-positive and metastasis-
negative nodes (Figure 10.1). Tumor-bearing nodes enhanced to a
greater degree and reached their peak value earlier than benign nodes.
Enhancement criteria produced higher sensitivity but lower specificity
for the diagnosis of metastasis than did unidimensional measurements
of a single node. The threshold enhancement value used for diagnosis was
a two-fold increase over baseline measured at 1 minute after injection. This
degree of enhancement approximated to 50 HU or 1.85 HU/g iodine
injected. Combined size–enhancement criteria produced even better
diagnostic performance, but a comparable performance could also be
achieved by a more complex morphological analysis that included the
size of adjacent axillary nodes.
120
No metastasis
Metastasis
Enhancement (HU)
80
40
0
0 2 4 6 8 10
−40
Time (min)
Figure 10.1 Mean ± standard deviation enhancement values in sentinel

axillary lymph nodes with and without metastases. (Adapted from reference 3)
9781842143094-Ch10 8/8/07 12:05 PM Page 166
Shim et al. adopted a different approach to the use of hemodynamic

studies for the prediction of nodal metastasis.4 Rather than measure
enhancement of the regional lymph nodes, these researchers measured
enhancement within the primary lung tumor itself (Figure 10.2).
Histological evidence for an association between intensity of angiogene-
sis and nodal metastasis in non-small-cell lung cancer provided the
basis for this methodology.5 Using a tube voltage of 120 kVp, volumetric
image acquisitions were performed every 20 s for 3 minutes following
a 36-g load of iodinated contrast material injected over 40 s. The study
group comprised 84 T1 tumors. Enhancement criteria produced higher
sensitivity but lower specificity for the diagnosis of metastasis than did
conventional morphological assessment of nodal status (Table 10.1). The
threshold enhancement value used for diagnosis was 60 HU, equivalent
to 1.67 HU/g iodine injected. Once again, a combination of size and
enhancement criteria produced the best diagnostic performance.
a b
c d
40
30
20
HU
10
0
1b 2b 3b 4b
−10
Time (s)
T1 n1
Figure 10.2 Conventional computed tomography (CT) (a) and maximal

enhancement image (b) of a T1 tumor non-small-cell lung cancer (arrow) with
enlarged ipsilateral hilar and mediastinal lymph nodes (chevrons). The time–
density curves following a 50-ml bolus of contrast material with iodine concen-
tration 300 mg/ml (c) demonstrate a maximum enhancement of 34 HU within
the primary tumor (red curve). This degree of enhancement is equivalent to
2.27 HU/g iodine and lies above the threshold of 1.67 HU/g that predicts nodal
metastasis. Hilar and mediastinal metastases are confirmed by fluorodeoxyglucose-
positron emission tomography (FDG-PET) (d). Note that the involved nodes
(green curve) demonstrate less enhancement than the primary tumor
9781842143094-Ch10 8/8/07 12:05 PM Page 167
40
Primary
Nodal mestastasis
30 Threshold
Size (mm) 20
10
0
0 Time
Figure 10.3 Theoretical growth curves for a primary tumor and nodal metastasis.
The double-headed green arrow demarcates a period of time in which enhance-
ment is readily measured in the primary tumor but the lymph node remains
normal by conventional size criteria
These three papers are interesting in that, despite different technical

approaches to different tumor types, the enhancement thresholds for
diagnosis of metastasis are similar, ranging between 1.67 and 2.5 HU/g
iodine injected. In each case, a combination of size and enhancement
criteria gave the best diagnostic performance. The use of enhancement
measures within the primary tumor, as adopted by Shim et al., holds
particular potential for the diagnosis of micrometastatic disease within
regional lymph nodes. The advantage of this approach is shown in
Figure 10.3, which illustrates theoretical tumor growth curves for a
primary tumor and nodal metastasis. Assuming that the primary tumor
must reach a certain size before metastasizing, the growth curve for the
nodal metastasis will lag behind that for the primary tumor. The hori-
zontal line demarcates the 10 mm size criterion typically used for the
diagnosis of metastasis. Note that there is a time period in which the
primary tumor is greater than 10 mm in size, and therefore suitable for
CT perfusion, yet the affected lymph node is still less than 10 mm in
size. This period of time offers an opportunity to use measurements of
enhancement within the primary tumor to detect nodal metastasis less
than 10 mm in size, i.e. when they are normal by size criteria and
difficult to measure reliably with CT perfusion.
LYMPHOMA
CT remains the mainstay imaging technique in the management of patients

with lymphoma, primarily for staging and assessment of treatment response.
9781842143094-Ch10 8/8/07 12:05 PM Page 168
However, there are a number of limitations to conventional CT imaging

in lymphoma, which perfusion imaging has the potential to address.
The choice of therapy for lymphoma is largely based on the grade
of tumor. The grade of lymphoma also carries important prognostic
significance. Although lymphoma grade is usually determined from his-
tological examination of lymph nodes, it is well recognized that tumor
grade can change from low to high during the natural history of the
disease.6 Furthermore, such transformation may occur with a subset of
the involved lymph nodes. Thus, in some cases it may be difficult to
confirm a change in grade without repeated and/or multiple biopsies.
The structural appearances of lymph nodes on CT provide little infor-
mation concerning tumor grade. However, studies by Ribatti et al.7 have
shown that microvessel density increases progressively from low- to
high-grade lymphoma. Although no direct correlation has been shown
for lymphoma, the correlation between microvessel density and CT per-
fusion measurements for other tumors suggests a potential role for CT
perfusion in assessing lymphoma grade.
A preliminary study by Dugdale et al.8 considered the relationship
between tumor grade and CT measurements of perfusion and permeability
in a mixed population of 39 patients with lymphoma. Of the 21 patients
who underwent perfusion measurement, five had low-grade and 16 had
intermediate/high-grade lymphoma. Perfusion measurements were higher
in patients with intermediate- or high-grade tumor than in those with
low-grade disease (56 ml/min/100 ml vs. 46 ml/min/100 ml). Although
there was overlap between patient groups, a perfusion value of greater
than 50 ml/min/100 ml suggested intermediate- or high-grade tumor
(Figure 10.4). Measurements of capillary permeability derived by
Patlak analysis in a separate group of 18 patients were unrelated to
lymphoma grade.
A further constraint to conventional CT in the assessment of patients
with lymphoma is that a residual tissue mass is a common finding
following treatment (Figure 10.5). A meta-analysis of the performance
of conventional CT in detecting active residual disease at completion of
chemotherapy, performed as part of a recent health technology assess-
ment in the UK, found low diagnostic specificity of only 45% (95%
confidence limits 27–64%).6 Based on decision modeling, this low
specificity would result in a CT-based strategy for the management of
patients with lymphoma on completion of therapy leading to 36% of
patients receiving unnecessary radiotherapy. Furthermore, the average
survival benefit and cost per patient for the CT-based strategy would be
only marginally better than a hypothetical management strategy in
which all patients received consolidation therapy without imaging.10
9781842143094-Ch10 8/8/07 12:05 PM Page 169
Figure 10.4 Conventional CT (a) and CT perfusion (b) of a patient with a

lymphoma of the small bowel mesentery prior to therapy. The perfusion value is
above 50 ml/min/100 ml implying high-grade disease
By reflecting angiogenesis, CT perfusion offers a means to assess the

activity of lymphoma and response to therapy on functional grounds.9
In view of the now recognized importance of angiogenesis in lym-
phoma, CT perfusion imaging offers a possible functional technique for
the assessment of therapeutic response, not only for conventional ther-
apy but also for the antiangiogenesis agents which are currently being
assessed as therapeutic agents for lymphoma. The potential for perfu-
sion imaging to be used as a marker for residual lymphoma activity is
further highlighted by the correlation between perfusion and glucose
metabolism within residual lymphoma masses that has been demon-
strated using [15O] water and FDG-PET.11
The preliminary functional CT study by Dugdale et al.8 also consid-
ered the potential for CT measurements of perfusion and permeability
9781842143094-Ch10 8/8/07 12:05 PM Page 170
a b
c d
Figure 10.5 Conventional CT (a, b) and corresponding CT perfusion images

(c, d) from two patients following treatment for lymphoma. The small nodal
masses in (a) and (c) (small arrows) demonstrate high perfusion implying
active disease, whereas the large mass (large arrow) in (b) and (d) shows low
perfusion implying inactive disease. The disease status was confirmed by FDG-PET.
(Adapted from reference 9)
to identify active residual disease (Figure 10.5). Median values of perfusion

were higher in patients with active disease (55 vs. 37 ml/min/100 ml),
and nodal perfusion below 20 ml/min/100 ml implied inactive disease
( p < 0.03). On the other hand, permeability values were little different
between these groups, and showed no correlation with change in activity
on serial studies.
When using CT perfusion as a tumor response marker (see Chapter 12), it
is important to be aware that certain treatments can induce direct effects
upon the vascularity of the tumor or adjacent tissues and so mask the
therapeutic response. For example, radiotherapy can cause a temporary
increase in CT values for both tumor perfusion and vascular permeability.12
9781842143094-Ch10 8/8/07 12:05 PM Page 171
Therefore, the timing of functional imaging for therapeutic monitoring

must take these factors into account.
SUMMARY
Although further studies are required to fully determine the clinical

value of the technique, studies performed to date highlight the poten-
tial for CT perfusion to be a useful adjunct to conventional structural CT
in the detection of lymph node metastases and the assessment of
disease grade and activity in patients with lymphoma.
REFERENCES
1. Guyatt GH, Lefcoe M, Walter S et al. Interobserver variation in the computed

tomographic evaluation of mediastinal lymph node size in patients with poten-
tially resectable lung cancer. Canadian Lung Oncology Group. Chest 1995; 107:
116–19.
2. Fukuya T, Honda H, Hayashi T et al. Lymph-node metastases: efficacy for
detection with helical CT in patients with gastric cancer. Radiology 1995; 197:
705–11.
3. Yuen S, Yamada K, Goto M, Sawai K, Nishimura T. CT-based evaluation of
axillary sentinel lymph node status in breast cancer: value of added contrast-
enhanced study. Acta Radiol 2004; 45: 730–7.
cancer enable the prediction of hilar or mediastinal nodal metastasis? AJR Am
J Roentgenol 2006; 186: 981–8.
5. Volm M, Koomagi R, Mattern J. PD-ECGF, bFGF, and VEGF expression in non-
small cell lung carcinomas and their association with lymph node metastasis.
Anticancer Res 1999; 19: 651–5.
6. Bastion Y, Sebban C, Berger F et al. Incidence, predictive factors, and outcome
of lymphoma transformation in follicular lymphoma patients. J Clin Oncol
1997; 15: 1587–94.
7. Ribatti D, Vacca A, Nico B et al. Angiogenesis spectrum in the stroma of B-cell
non-Hodgkin’s lymphomas. An immunohistochemical and ultrastructural
study. Eur J Haematol 1996; 56: 45–53.
8. Dugdale PE, Miles KA, Kelley BB, Bunce IH, Leggett DAC. CT measurements
of perfusion and permeability within lymphoma masses: relationship to grade,
activity and chemotherapeutic response. J Comput Tomogr 1999; 23: 540–7.
9. Miles KA. Lymphoma and myeloma: monitoring treatment response with
[18F]FDG-PET and other functional imaging techniques. ICIS Cancer Imaging
Volume 4: S131–5 (posted 3 January 2005).
10. Bradbury I, Bonell E, Boynton J et al. Positron emission tomography (PET)
imaging in cancer management. Health Technology Assessment Report 2.
Glasgow: Health Technology Board for Scotland, 2002.
9781842143094-Ch10 8/8/07 12:05 PM Page 172
11. Dimitrakopoulou-Strauss A, Strauss LG, Goldschmidt H et al. Evaluation of

tumor metabolism and multidrug resistance in patients with treated malignant
lymphomas. Eur J Nucl Med 1995; 22: 434–42.
12. Harvey CJ, Blomley MJ, Dawson P et al. Functional CT imaging of the acute
hyperemic response to radiation therapy of the prostate gland: early experience.
9781842143094-Ch11 8/8/07 4:25 PM Page 173
11
CT perfusion of liver
metastases and early detection
of micrometastases
Charles A Cuenod, Laure Fournier, Daniel Balvay, and Kenneth A Miles
INTRODUCTION
The liver is the second site of metastatic disease,1 and the most
common site of metastasis of colorectal carcinoma.2 Eighty percent of
patients with colorectal carcinoma recurrence die within 3 years, pre-
dominantly from hepatic metastases.3
Despite the availability of various therapeutic options, hepatic metas-
tases remain difficult to eradicate, partly due to their late discovery.
Indeed, imaging techniques such as magnetic resonance imaging (MRI),
computed tomography (CT), and ultrasound (US) allow diagnosis of
most liver tumors over 1 cm (sensitivity of 55–90%),4 but, for smaller
lesions, sensitivity is much lower (under 50%), and microscopic lesions
remain occult. Liver metastases are present but not detected in up to
one-third of patients who undergo apparently curative excision of
primary colorectal carcinoma.5
Therefore, the identification of patients with micrometastatic disease
who are at risk of recurrence would have profound implications for
prognosis and treatment, since it would allow the implementation of
adjuvant chemotherapy at an earlier stage.6 On the other hand, exclu-
sion of micrometastastic disease could obviate potentially morbid
chemotherapy in patients without seeds of micrometastasis.
Hopes for the earlier detection of liver metastases have been raised
by functional imaging suggesting that occult metastases induce changes
in liver blood flow similar to those caused by overt metastases,7–11
such as a decrease in the portal perfusion of the liver and an increase
9781842143094-Ch11 8/8/07 4:25 PM Page 174
in arterial perfusion. CT perfusion, therefore, could become a con-

venient and easy way for early detection of occult liver metastases.
In this chapter, we will review the normal microcirculation of the
liver and the methods for its quantitative analysis by functional CT.
Changes in liver microcirculation induced by the presence of overt and
occult metastases will be presented and discussed.
LIVER PERFUSION PHYSIOLOGY
The liver has a unique perfusion system with a dual blood supply. More
than two-thirds of the blood supply comes from the low-pressure
portal vein, and the rest comes from the high-pressure hepatic artery
(Figure 11.1). The capillaries of the liver, called ‘sinusoid capillaries’,
therefore contain a mixture of portal and arterial blood.
The sinusoid capillary system is very dense, representing almost one-
third of the volume of the liver parenchyma (whereas in the brain, for
example, the blood within the capillaries contributes to only 4%
of the volume of tissue). The endothelial cells that line the sinusoids
are anatomically and biologically distinct from endothelial cells in
Hepatic veins
output C°(t)
Hepatic artery Ca(t)

CT(t)
input 1
Liver parenchyma Slice
Portal vein
input 2
Cp(t)
Figure 11.1 The liver is supplied more than two-thirds by the portal vein and
less than a third by the hepatic artery. A transverse slice through the abdomen
allows simultaneous measurement of the attenuation in regions of interst (ROIs)
drawn in the portal vein, in the aorta, and in the liver tissue
9781842143094-Ch11 8/8/07 4:25 PM Page 175
CT perfusion of liver metastases 175
other organs. They lack a basement membrane and contain fenestrae,

characteristics that facilitate the transport of nutrients and macromole-
cules between the sinusoids and the hepatic parenchyma.
The hepatocytes lining the sinusoids represent most of the volume
of the liver parenchyma, and the interstitial space, called the ‘space
of Disse’, is almost virtual, representing only 10% of the total liver
volume. Specialized pericytes called stellate cells (or Ito cells) are
located in the space of Disse and wrap around the walls of the hepatic
sinusoids.
Regulation of blood flow and vascular resistance in the hepatic
microvasculature are intimately linked. Whereas in most organs the site
of blood flow regulation occurs at the arteriolar level, in the liver most
of the blood flow enters at low pressure through the portal vein, and
resistance changes occur in the sinusoid. Stellate cells play a role in the
regulation and control of blood flow through the liver based on their
anatomic location and contractile characteristics by modulating the
sinusoidal caliber in response to several vasoactive endothelium-
derived mediators including nitric oxide (NO).12,13
The portal supply varies greatly during the day in relation to bowel
activity, with large increases in the postprandial periods. The total
hepatic blood supply, however, is finely tuned by the so-called ‘hepatic
arterial buffer response’, which is the inverse response of the hepatic
artery to changes in portal vein flow.14 These intrinsic regulatory mech-
anisms based on the local concentration of adenosine tend to maintain
total hepatic blood flow at a constant level, allowing an increase in the
arterial blood supply to compensate for a decrease in portal supply, and
a decrease in arterial blood supply in cases of increased portal supply.15
It is important to note that, in contrast, variations in arterial blood
supply cannot be compensated by variations in portal supply.
QUANTITATIVE MEASUREMENT OF LIVER

MICROCIRCULATION WITH FUNCTIONAL CT
Introduction
The specific dual perfusion of the liver makes it more difficult to ana-
lyze with contrast-enhanced imaging than the perfusion of other tissues.
The enhancement curve of the liver, after the injection of a bolus of
contrast agent, is the combination of the enhancement due to the con-
trast agent flowing in the arterial blood and the contrast agent flowing
in the portal blood. Whereas the molecules of contrast agent arrive
9781842143094-Ch11 8/8/07 4:25 PM Page 176
quickly when they are delivered to the liver through the arterial route,
they are delayed and diluted by the splanchnic circulation when they
are delivered through the portal route.
Many strategies have been developed to take advantage of the portal
lag to separate and quantify the arterial and portal hepatic perfusions.
A comprehensive review of perfusion imaging of the liver has recently
been published by Pandharipande et al.16
Different methods for liver perfusion quantification with CT

Slope-ratio methods
Miles et al.17,18 described liver perfusion imaging using CT in 1993 by
generating enhancement curves from regions of interest (ROIs) drawn
over the liver, the aorta, and the spleen after a bolus injection of con-
trast agent. Liver enhancement was resolved into arterial and portal
venous components by assuming that maximum splenic enhancement
marks the end of the early arterial phase and the beginning of the
delayed portal venous phase of liver perfusion. The maximal slopes of
the liver time–density curve in each phase were divided by the peak
aortic enhancement to calculate both arterial and portal perfusion
(Fa, Fp). The hepatic perfusion index (HPI), which is the ratio of the
arterial perfusion to the total hepatic perfusion (HPI = arterial perfu-
sion/arterial + portal perfusion) was also calculated. HPI is also known
as the ‘Hepatic Arterial Fraction’, see Chapter 2.
This technique is simple to implement and can be applied to any
segment of the liver, as there is no need to include the portal vein or
major portal vessel within the tissue imaged. However, the method
underestimates portal hepatic flow for two reasons: first, the down-
wards slope of the last part of the arterial time–attenuation curve is
superimposed on the upwards slope of the arriving portal curve; and
second, the maximal slope of the portal venous phase of enhancement
is divided by the peak aortic enhancement instead of the peak portal
enhancement, which is flattened and diluted after flowing through the
splanchnic system.
To avoid these limitations, Blomley et al.19,20 modified this approach
by subtracting the arterial phase liver enhancement (modeled after
splenic enhancement) from the liver enhancement curve to give a more
‘accurate’ portal time–attenuation curve. From this corrected curve,
portal perfusion was calculated by dividing the slope of the rise in
attenuation during the portal enhancement phase by peak portal
venous enhancement itself.
9781842143094-Ch11 8/8/07 4:25 PM Page 177
This ‘corrected approach’, however, has two main limitations. First,

it assumes that hepatic arterial and splenic enhancement curves are
similar. Such similarity is unlikely in view of the unique microcircula-
tion within the spleen, recognized as the mechanism underlying the
transient splenic inhomogeneity seen during contrast-enhanced CT.21
Second, the technique requires a set of slices containing both the portal
vein and a part of the spleen to be able to draw ROIs and extract the
enhancement curves.
The slope-ratio methods have largely been used for the evaluation
of hepatic perfusion in metastatic disease.18,22,23 Since these methods
take into account only the rising part of the liver enhancement curve,
before the venous outflow, they do not give information regarding the
hepatic blood volume (HBV) and the mean transit time (MTT) through
the sinusoid network. A sharp bolus of contrast medium is essential to
avoid underestimation of the arterial and portal perfusion from venous
outflow of contrast, which can be considered negligible if the outflow
does not begin before the end of the input.
Tracer kinetic modeling

To avoid the approximations of the arterial and portal slopes and take
into account more of the liver enhancement dynamics, tracer kinetic
modeling techniques have been developed24–28
Compartmental model Van Beers and Materne have developed a

compartmental model with a dual-input:25,26
dCL(t)/dt = k1aCa(t) + k1pCp(t) − k2CL(t)
in which CL, Ca, and Cp are the contrast agent concentrations measured
over time respectively within the liver, hepatic artery, and portal vein
derived from ROIs, and k1a, k1p, and k2 are the arterial and portal
venous inflow and liver outflow rate constants. By fitting measured
CL(t), the constants k1a, k1p, and k2 can be estimated, and can be used
to calculate hepatic arterial and portal venous perfusion, mean transit
time (MTT) of contrast agent through the liver, and contrast agent dis-
tribution volume within the liver (Vd). The distribution volume Vd of
contrast agent is used instead of the hepatic blood volume, because the
small-molecule contrast agent used in CT leaks freely and instanta-
neously across the sinusoid capillary wall, leading to a distribution
volume that associates the hepatic blood volume and part of the extra-
cellular Disse space.
9781842143094-Ch11 8/8/07 4:25 PM Page 178
This compartmental approach assumes that there is an instantaneous

mixing of blood in the capillary compartment.
Deconvolution model To avoid this assumption in the liver, where

the capillary network is complex and the mean transit time long, Cuenod
et al. have developed27–29 a specific deconvolution technique. The decon-
volution method considers that the time course of the concentration of
contrast agent entering a tissue is modulated by a transfer function spe-
cific to the tissue. This transfer function can be computed from both the
concentration–time curve of contrast agent entering the tissue (input)
and the concentration–time curve of contrast in the tissue. From that
transfer function, the perfusion parameters of the tissue can be calcu-
lated. The deconvolution strategy was introduced into functional CT by
Axel in the 1980s.30 The specificity of the liver, for this approach, comes
from its dual vascular input. The method is described below.
Deconvolution method for liver perfusion imaging

Theory of the deconvolution method (see also Chapter 2)
Deconvolution allows the determination of the theoretical impulse
response of the tissue, that is, the time course of concentration that an
instantaneous input of contrast material (impulse input) would have
yielded.
Convolution When the contrast agent enters the tissue as a function

of time, Ci(t), the time course of the concentration of contrast through-
out the tissue depends both on the time course of a theoretical impulse
input (instantaneous input) through the tissue, h(t), and on the actual
experimental time course of contrast input.30 The concentration–time
curve at the venous outflow, Co(t), is the convolution of Ci(t) by h(t):
Co(t) = Ci(t) ƒ h(t)
Residue function However, since we cannot measure the concentra-

tion–time curve at the venous outflow of the tissue and can only meas-
ure the concentration–time curve of contrast into the tissue, we have to
infer the venous outflow concentration–time curve from the tissue con-
centration–time curve. To do so, we use the notion of residue function.
The integral of h(t) is:
H (t ) = ∫ h (τ ) d τ
t
0
9781842143094-Ch11 8/8/07 4:25 PM Page 179
h(t) = dH(t)/dt = -dR(t)/dt Probability density function
h (t)
1.0
H (t) H(t)= h(t)dt
Cumulative frequency function
1.0
Residue function
R (t) R(t)=1-H(t)
Figure 11.2 Relations between the residue function R(t), the cumulative
frequency function H(t), and the probability density function h(t). The contrast
agent that progressively leaves the tissue accumulates outside the tissue. At the
venous outlet the concentration of contrast agent rises progressively before
decreasing to zero. The initial value of R(0) is normalized to one, as well, there-
fore, as the final value of H(ⴥ) and the area under the curve h(t). (Adapted
from reference 31)
where H(t) is the fraction of an impulsive input which has already

left the tissue by time t (Figure 11.2). It is called the cumulative
frequency function. Its complementary function is called the residue
function R(t):
R(t) = 1 − H(t)
where R(t) is the fraction of the impulsive input remaining within its
distribution volume Vd in the tissue at time t.31
The concentration–time curve of a tracer remaining in its volume of
distribution within the tissue Cd(t) can be predicted for any type of
input function, Ci(t), as the convolution of Ci(t) by R(t):
Cd(t) = Ci(t) ƒ R(t)
Because the distribution volume Vd of the tracer is within a larger

volume of tissue Vt, the concentration of tracer within the tissue is:
Ct(t) = Cd(t)⋅Vdt
9781842143094-Ch11 8/8/07 4:25 PM Page 180
where Vdt = Vd/Vt is the fractional dilution volume of the molecule

expressed as a percentage of the total volume of tissue Vt. The concen-
tration of tracer in a voxel of tissue Ct(t) is therefore the convolution of
Ci(t) by Rt(t), the tissue transfer function (Rt(t) = [Vdt ⋅ R(t)]):
Ct(t) = Ci(t) ƒ Rt(t)
Estimation of Rt(t), R(t), and h(t) After contrast injection, a decon-

volution process can allow the determination of Rt(t), knowing the con-
trast variation of the tissue Ct(t) and the input function Ci(t). For finite
time sampling steps of ∆t = T, the convolution Ct(t) = Ci(t) ƒ Rt(t) can
be approximated by the following sum:
Ci ( nT ) ⊗ R t ( nT ) = T ∑ kk == 0n−1 Ci ( nT − kT ) ⋅ R t ( kT )
The category of Weibull functions g ( t ) = a ⋅ exp [ − t/b c ] has been

chosen to represent the tissue transfer function Rt(t) because its shape
is intermediate between a falling exponential function and a square
function, resembling the supposed liver curve.
A computer program is necessary to minimize the quadratic error
between the measured tissue response Ct(nT) at each time nT and the
assumed response C*(nT)
t after convolution of the measured input
Ci(nT) at each time nT:
⎡ ⎛ kT ⎞ c ⎤
C∗t ( nT ) = T ∑ kk==0n−1 ⎡⎣Ci ( nT − kT )⎤⎦ ⋅ a ⋅ exp ⎢− ⎜ ⎟ ⎥
⎢⎣ ⎝ b ⎠ ⎥⎦
The program yields the value of the three unknown factors a, b, and c,
allowing the estimation of Rt(t).
Since Rt(t) = Vd ⋅ R(t) and R(0) = 1, then Vd = Rt(0) and R(t) =
Rt(t)/Rt(0).
When R(t) has been worked out, h(t) can be obtained as its negative
derivative
dR ( t )
h(t) = −
dt
and the output function Co(t) can be calculated.

9781842143094-Ch11 8/8/07 4:25 PM Page 181
Extraction of microvascular parameters

Measurement of the mean transit time The mean transit time
(MTT) through the vascular bed can be calculated as the first moment
(or the geometric mean) of the calculated Co(t):
∫ tC ( t ) dt
∞
0
MTT = 0
∫ C ( t ) dt
∞
0 0
It can also be calculated as the first moment of the impulse response

itself:
∫ th ( t ) dt
∞
MTT = 0
∫ h ( t ) dt
∞
∫ h ( t ) dt = 1 ∫ th ( t ) dt.
∞ ∞
and even more simply, knowing that, as MTT =
0 0
Measurement of the fractional volume of distribution The frac-

tional distribution volume of the tracer Vdt, can be calculated as Rt(0),
the initial (maximal) value of Rt(t).
As expressed above, the fractional distribution volume of the con-
trast within the tissue, Vdt, is calculated as the initial (maximal) value of
Rt(t): Vd = Rt(0).
Measurement of the tissue blood flow with the central volume

theorem The blood flow through a unit volume of tissue (Ft =
F/volume of the organ) expressed as ml/min/100 ml is measured using
the central volume theorem:
Vdt
Ft =
MTT
Extraction of hepatic perfusion parameters

Specifically in the liver, the dual blood supply has to be taken into
account for the calculation (Figure 11.1) The respective balance
between the arterial input Ca(t) and venous portal input Cp(t) of the
liver is expressed as the hepatic perfusion index (HPI), which is the
ratio of the arterial blood flow (Fa) over the total hepatic blood flow Ft:
Fa
HPI =
Fa + Fp
9781842143094-Ch11 8/8/07 4:25 PM Page 182
In the liver, arterial and portal blood are mixed in the sinusoidal capil-
laries, and the tissue concentration–time curve in the liver (referred to
as CL) can be expressed as:
C t ( t ) = ⎡⎣α Ca ( t ) + (1 − α ) Cp ( t ) ⎤⎦ ⊗ R t ( t )
The computer program has therefore to minimize the quadratic error

between the actual tissue response Ct(nT) and the assumed response
C *(nT)
t after convolution of the dual input (Figure 11.3):
⎡ ⎛ kT ⎞ ⎤
( )
c *t ( nT ) = T ∑ kk == 0n=1 αC a ( nT − kT ) + (1 − α ) Cp ( nT − kT ) ⋅ a ⋅ exp ⎢− ⎜ ⎟ c ⎥
⎣ ⎝ b⎠ ⎦
The program yields the value of the four unknown factors α, a, b, and
c, allowing the estimation of Rt(t) and α = HPI. Rt(t) allows the
calculation of MTT, Vdt, and FT, and HPI allows the calculation of
Fa = HPI × FT, and Fp = (1 − HPI) × FT.
80
60
∆HU
40
20
0
0 20 40 60 80 100
Time (s)
Figure 11.3 The time–enhancement curve of the liver, expressed as Hounsfield

unit variation (∆HU) over time (continuous line), can be separated by the
computer using the deconvolution model into the linear combination of
the early and small enhancement curve of the arterial supply (crosses), and the
late and strong enhancement curve of the portal supply (stars). The contrast
enhancement is obtained by subtracting the mean baseline value from the
values measured in the ROIs
9781842143094-Ch11 8/8/07 4:25 PM Page 183
TABLE 11.1 The six main liver perfusion parameters that can be extracted with
functional computed tomography (CT)
Name Abbreviation Definition Unit
Mean transit time MTT Mean time taken by s

molecules of contrast
agent to flow through
system
Liver distribution LDV Percentage of tissue volume % or ml/100 ml of tissue
volume in which the contrast
agent distributes itself
Total hepatic FT Total hepatic blood flow ml/min/ml of tissue
blood flow FT = Fa + Fp
Arterial blood flow Fa Hepatic blood flow of ml/min/ml of tissue
arterial origin
Portal blood flow Fp Hepatic blood flow of ml/min/ml of tissue
portal origin
Hepatic perfusion HPI Percentage of total blood %
index flow of arterial origin
Fa
HPI =
Fa + Fp
These parameters are obtained by drawing regions of interest (ROIs)

on the aorta, the portal vein, and the liver parenchyma. The liver’s ROI
has to be drawn as large as possible, avoiding the large vessels. Then,
the three ROIs are replicated by the computer on each image of the
series to extract the CT attenuation numbers (expressed as Hounsfield
units) over time. The time–attenuation curves derived from the aorta
Ca(t), the portal vein Cp(t), and the liver Ct(t) can then be used for
calculation of the six hepatic perfusion parameters (Table 11.1).
When the analysis is run on a pixel-by-pixel basis, it yields para-
metric maps displaying the value of each parameter in each pixel
(Figure 11.4).
FUNCTIONAL DETECTION OF LIVER METASTASES
CT perfusion and overt metastases

In patients with known metastatic disease, increased arterial perfusion
has been shown by Miles et al.18,23,32 and Blomley et al.19 with values
around 40–50 ml/min/100 ml in patients with known metastatic disease
versus values of 17–19 ml/min/100 ml in healthy control groups
(Figure 11.5). Leggett et al.22 also demonstrated increased arterial
9781842143094-Ch11 8/8/07 4:25 PM Page 184
Transit time Permeability

50
40
30
HU
20
10
0
0 10 20 30 40
−10
time (s)
Iiver
Perfusion Blood volume HPI
Figure 11.4 By analysis of the liver time–density curve (top left), it is possible
to derive parametric images displaying a range of parameters reflecting the
hepatic vasculature. (HPI; hepatic perfusion index)
a b
c d
Figure 11.5 Conventional portal phase computed tomography (CT) (a) and CT
perfusion images of a hepatic metastasis from colorectal cancer. The metastasis
exhibits increased arterial perfusion (b) and hepatic perfusion index (c) but low
levels of portal perfusion (d). Note that the area of increased arterial perfusion
and hepatic perfusion index extends beyond the margins of the metastasis
9781842143094-Ch11 8/8/07 4:25 PM Page 185
TABLE 11.2 Perfusion parameter values (median [range]) in rats with micrometastases
and macrometastases compared to control rats
Micrometastases
Heamodynamic Normal livers normal appearance
parameters control rats of liver Macrometastases
HPI (%) 17 [19] 9.2 [23] 74.6 [70]**

MTT (s) 7.93 [4] 9.94 [4.96]* 25.19 [16.17]**
LDV (%) 46.4 [7.8] 43.5 [13.9] 29.5 [14.5]**
FT (ml/min/ml) 3.23 [1.9] 2.07 [0.94]* 0.71 [0.71]**
Fa (ml/min/ml) 0.5 [0.7] 0.21 [0.44] 0.41 [0.25]
Fp (ml/min/ml) 3.08 [2.4] 2.04 [1.28]* 0.20 [0.78]**
(Mann–Whitney U test, *p < 0.05, **p < 0.015) (Data from references 27 and 28)
perfusion (> 25 ml/min/100 ml) in nine of 11 patients with overt

colorectal metastases examined using the same CT perfusion technique,
and a decrease in portal liver perfusion in five of these patients.
In the rat, Cuenod et al., using the deconvolution technique, found
in visible liver metastases from colon cancer, a large increase in HPI and
a decrease in liver perfusion due to a decrease in portal perfusion.27
A decrease in the distribution volume and an increase in MTT were also
observed (Table 11.2).
Moreover, Tsushima et al.33 showed that patients with liver metas-
tases had abnormal blood flow in apparently normal liver compared to
controls, and this difference was not seen in subjects with malignancy
but without liver metastases.
Micrometastases
Modifications of liver perfusion can be found not only in patients with
overt liver metastases, but also in patients bearing occult microscopic
disease who reveal liver metastases on follow-up.
In a cohort of 80 patients with colon cancer and without overt liver
metastases, Platt et al.11 showed, on classic liver CT examination, that
liver enhancement 40 seconds post-injection was higher in 22 patients
who revealed liver metastases in the following 18-month period. With
a threshold of 0.4 for the ratio of the difference between the liver den-
sity before injection and that at 40 s over the maximum liver enhance-
ment, the test had a sensitivity of 75% and a specificity of 96%.
9781842143094-Ch11 8/8/07 4:25 PM Page 186
Leggett et al.22 in addition to describing liver perfusion changes in

patients with overt metastases as mentioned earlier, also showed a
decrease in portal liver perfusion in the three patients who revealed
hepatic metastases during the follow-up period, among 16 patients
without initial liver metastases.
In animals, similar results have been reported by Cuenod et al.27
showing decreased portal perfusion and an increased HPI associated
with an increased mean transit time in the livers of rats bearing occult
metastases.
Relationship between hepatic CT perfusion and survival

Cancer patients entering imaging surveillance programs following suc-
cessful primary treatment do not represent a uniform population of
equal risk of recurrence. The identification of predictive factors that are
linked to outcomes may allow the modification of surveillance strate-
gies for subgroups of patients, and the need for research in this area
has been highlighted by an expert panel of the American Society of
Clinical Oncology.34 Although several laboratory-derived predictive
factors have been identified for patients with colorectal cancer,35,36 rel-
atively little attention has been given to biomarkers derived from diag-
nostic imaging. CT perfusion represents one of the recent developments
in physiological imaging that can provide new opportunities for the use
of imaging as a biomarker.37
There is preliminary evidence to suggest that CT measurements of
hepatic perfusion may relate to patient survival, not only in the pres-
ence of visible metastases but also when no overt lesions are depicted
by conventional CT (Figure 11.6). In a series of 13 patients with hepatic
metastases from a range of primary tumors, Miles et al. found that
survival of the patient correlated significantly with arterial perfusion
values in the metastasis and adjacent liver.23 A similar association has
also been reported from histological studies of tumor angiogenesis in
metastatic colorectal cancer.38 In a separate study, the relationship
between hepatic CT perfusion and survival was evaluated in a larger
series of 80 patients with colorectal cancer.39 Patients were stratified on
the basis of conventional and CT perfusion as follows: (1) no visible
metastases on CT and HPI < 0.35, (2) no visible metastases on CT and
HPI ≥ 0.35 and (c) overt liver metastases on CT. No metastases were
visible in 36 patients, and distribution of the modified Dukes classification
was A: 5, B: 9, C: 48, D: 18. Stratification of survival risk by CT
perfusion appeared to be superior to that based on the Dukes classifi-
cation (Figure 11.7).
9781842143094-Ch11 8/8/07 4:25 PM Page 187
> 46 months 31 months 9 months
a 40 b 40 c 40
Enhancement (HU)
Enhancement (HU)
Enhancement (HU)
30 30 30
20 20 20
10 10 10
0 0 0
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Time (s) Time (s) Time (s)
Figure 11.6 Three cases illustrating the relationship between hepatic enhance-
ment and survival. Conventional CT images are displayed above with corre-
sponding time–density curves (TDCs) below. The arrows indicate the separation
of arterial and portal phases of enhancement as identified by the time of peak
splenic enhancement. The first case (a) survived for more than 46 months fol-
lowing CT, which demonstrated no visible metastases on conventional images.
The hepatic TDC is normal with greater enhancement in the portal phase. The
second case (b) also had no visible metastases on CT but survived for a shorter
time (31 months). The hepatic TDC shows increased enhancement in the
arterial phase implying possible micrometastatic disease. The third case (c) had
visible metastases and survived only 9 months following CT. Arterial phase
enhancement is considerably raised
a 0.1 b 0.1
0.8 0.8
Survival
Survival
0.6 0.6
0.4 0.4
HPI < 0.35 no Mets Dukes A
0.2 HPI < 0.35 + Mets 0.2 Dukes B
HPI ≥ 0.35 no Mets Dukes C
HPI ≥ 0.35 + Mets Dukes D
0 0
0 12 24 36 0 12 24 36
Time (months) Time (months)
Figure 11.7 Kaplan–Meier curves for a cohort of 80 patients with colorectal

cancer stratified by perfusion CT (a) and Dukes classification (b). The stratifi-
cation is superior with perfusion CT (blue = no visible metastases (Mets) and
HPI < 0.35, red = no visible metastases and HPI ≥ 0.35, yellow = visible metas-
tases and HPI ≥ 0.35, green = visible metastases and HPI < 0.35). (Data from
reference 39)
9781842143094-Ch11 8/8/07 4:25 PM Page 188
Other imaging techniques

The results found in CT are validated by similar results found previously
with functional scintigraphy and duplex Doppler ultrasound, and more
recently with MRI. Furthermore, hepatic CT perfusion parameters have
been shown to correlate with dynamic liver scintigraphy17 and Doppler
ultrasound.40
Dynamic liver scintigraphy

It was shown as early as 1987 by quantitative hepatic scintiangiography
that the presence of liver metastases induces changes in hepatic arterial
and portal venous blood flows in animals and in humans.7–9 Although
values of hepatic arterial and portal venous blood flow were difficult to
measure and the reproducibility of the computed parameters was quite
poor, HPI was achievable, and was shown to increase not only in the
presence of overt metastases9 (above a HPI threshold value of 0.37,
metastatic disease should be suspected), but also in patients for whom
liver metastases appeared subsequently during follow-up,7,41,42 as well
as in animals bearing occult metastases.
Ballantyne and Cartert pointed out that a single estimation of HPI
was not reliable for the identification of patients with overt hepatic
metastases,41,43 and suggested, therefore, the use of the rise of HPI on
serial studies, which is associated with progression of disease and
increases the reliability of the test.
Duplex Doppler ultrasound

Similar results have been shown using Duplex Doppler ultrasound.10,44,45
In this technique, the blood flows reaching the liver though the portal
vein and the hepatic artery are calculated from the product of the veloc-
ities and the cross-sectional areas of the respective vessels. Although
there is no significant difference in total liver blood flow in patients
with colorectal liver metastases when compared to controls, the ratio of
the hepatic arterial flow to the sum of the portal and hepatic arterial
flows (Doppler perfusion index, DPI, which is close to HPI for the
entire liver) is markedly elevated in patients with overt metastases and
elevated in patients with occult metastases. For example, Leen et al.46
showed that the 5-year survival of patients after potentially curative
colon surgery for colorectal carcinoma was 91% in patients with
normal hepatic Doppler perfusion index values and 29% in patients
9781842143094-Ch11 8/8/07 4:25 PM Page 189
with abnormal hepatic DPI values. The technique, however, depends

highly on the skills of the investigators, and is difficult to replicate,47
with overlapping values in groups of patients with and without overt
metastases.
In an experimental animal model of hepatic metastases, Yarmenitis
et al.48 demonstrated that metastases were first encountered on day 4 as
small groups of cells in the connective tissue of the porta hepatis and
the portal triads, without apparent vascular association, and were asso-
ciated with a clear elevation of arterial flow and DPI and a subtle reduc-
tion in portal flow.
Contrast-enhanced dynamic MRI

More recently, Totman et al.49 showed, by using contrast-enhanced MRI
with Gd-DTPA (gadopentetic acid), that patients with overt colorectal
liver metastases had higher HPI than that of normal controls, but they
did not test the method on occult liver metastases.
PHYSIOPATHOLOGY OF MICROCIRCULATION
ALTERATIONS IN METASTATIC LIVER
All the imaging techniques have shown similar flow alterations in liver
bearing visible metastases. The main finding is an increase in HPI (or
DPI with Doppler), which appears to be mostly due to a decrease in
portal perfusion, and to a lesser extent to an increase in arterial perfu-
sion. Such a relative increase in arterial supply in the environment of
liver metastases had already been described in 1954.50 An increase in
MTT and a decrease in distribution volume are also found.
Surprisingly, significant intrahepatic flow alterations also occur prior
to the growth of visible metastases. These hemodynamic changes are
similar but more discrete than those found in the presence of large
metastases.
The precise cause of these changes, however, remains only partly
explained. Both biomechanical and molecular mechanisms appear to
be implicated.
Humoral mediators have been implicated.51 The presence of circu-
lating vasoactive agents has been suggested by changes induced in
splanchnic hemodynamics in rats by injecting plasma samples from
patients or rats bearing tumors,52,53 but the nature of these mediators is
currently not known.
9781842143094-Ch11 8/8/07 4:25 PM Page 190
In vivo videomicroscopy
Direct observation using in vivo videomicroscopic techniques has made
it possible to study in live animals the flow alterations in hepatic metas-
tases and in the liver during their formation.54–57
Large tumor nodules contain irregularly dilated, tortuous, neoangio-
genic vascular networks. Blood flow velocity and direction vary in this
network, accounting for the increase of the mean transit time. The nod-
ules induce direct mechanical compression on the surrounding sinu-
soids, which are narrowed, and resist portal low-pressure flow. The
portal venules and the sinusoids are obliterated abruptly at the borders
of the growing metastatic foci,55 and most flow is stopped, accounting
for the drop in portal flow within the tumor and for the portal hyper-
tension in tumor-bearing liver. The drop in portal flow is compensated
by arterial flow58 via the arterial buffer effect, accounting for the
increase in HPI as well as the peritumoral rim and the transient segmen-
tal enhancements that can be observed during the arterial phase on CT
and MRI.59
Moreover, although metastases smaller than 200 µm receive their
blood supply via the sinusoids, without hepatic arterial neovasculariza-
tion,60 when tumors progress to a point beyond which angiogenesis is
necessary to sustain growth, neovascularization occurs,61 assisted by
vascular endothelial growth factor expression.62
Since significant intrahepatic flow alterations occur in livers before
metastases become visible, mechanisms other than extrinsic compres-
sion of sinusoids must be implicated. Intravital microscopy during
the formation of hepatic metastases showed that several intrahepatic
microhemodynamic events occur prior to the establishment of visible
metastases:57
• blocked metastatic cells within the sinusoids56
• increased post-sinusoidal leukocyte rolling and adherence, and stasis
of leukocytes in the sinusoids
• reduced wall shear rate (velocity gradient perpendicular to the direc-
tion of fluid flow)
• and, finally, a reduction in blood flow in the hepatic sinusoids.
Metastatic cells coming through the portal route are arrested at
the inflow side of the microvascular bed due to size restriction.56
Microthrombi form in the portal vessels55,60 and increase resistance in
the low-pressure vascular bed, which accounts for the increased MTT
and the decreased portal flow. Kruskal et al.57 showed that when com-
pared with normal livers, a reduction in sinusoidal flow velocity and
9781842143094-Ch11 8/8/07 4:25 PM Page 191
flow rate began as early as day 2 after the injection of tumor cells in the
portal system of mice that developed hepatic metastases, and they
noted a significant decrease in post-sinusoidal leukocyte velocity and
an increase in the number of stagnant adherent leukocytes, associated
with a decrease in wall shear rate. To resist substantial wall shear stress
exerted by blood flow, metastasizing cells have to form adhesive
contacts with endothelial cells. The reduced shear rate increases the
likelihood of interactions between leukocytes (and tumor cells) and
endothelial receptors, favoring formation of metastases. In return,
increased leukocyte rolling and adherence narrow the sinusoids and
portal venules and compromise flow.
Moreover, cell membrane molecules such as the carcinoembryonic
antigen, present in certain cancer cells including colon cancer cells,
bind to Kupffer cells by specific receptors, and activate them. Activated
Kupffer cells can release vasoactive cytokines that activate stellate cells,
which impair sinusoidal perfusion by modulating the sinusoidal caliber.
The increase in high-pressure arterial flow which partly compensates
for the portal flow decrease may be due to the buffer effect, or to
humoral factors also secreted by tumoral cells.
The low distribution volume of the contrast medium within the
tumor suggests that the capillary density is lower than in the normal
liver, and accounts for the fact that metastases from colon and breast
cancers usually appear to be less enhanced than normal liver after
contrast medium injection.55
LIMITATIONS OF CT PERFUSION IN THE LIVER
CT perfusion imaging of the liver has the potential to improve detection

and characterization of liver diseases; however, several challenges remain.
The main limitations result from motion in the abdomen.63 Although
the arterial input function is easily measurable in the abdominal aorta,
the portal input function is much more difficult to obtain due to breath-
ing artifacts. A vessel tracking strategy, using imaging processing tech-
niques, can greatly improve the shape of the portal enhancement curve
and therefore the model adjustment (unpublished data). Respiratory
motion also makes it difficult to analyze nodules in the liver. Two
strategies can be applied to avoid this limitation: respiration gating and
compensation for respiratory misregistration.64
Respiration gating, however, implies a low acquisition rate, which
may compromise the separation of the arterial and the portal contribu-
tions to the liver perfusion.
9781842143094-Ch11 8/8/07 4:25 PM Page 192
Retrospective motion correction is difficult to achieve in the liver due

to elastic deformation of the organ during the respiration cycle.
Therefore, rigid three-dimensional (3D) motion correction algorithms,
which are very efficient in the brain, must be replaced by more com-
plex, non-rigid algorithms. These will be easier to run in large volumes
acquired with millimetric slices obtained with the newer multidetector
CT systems.
Another limitation comes from the high number of parameters that
are required to model the dual liver microcirculation if the capillary leak
is to be calculated in tumor nodules. In liver perfusion models, meas-
urements are performed during the first pass of the contrast (avoiding
the capillary leak), or are based on the hypothesis that the leak in the
interstitial space of Disse is almost immediate due to fenestration of the
sinusoids (giving a hepatic distribution volume instead of a hepatic
blood volume), or that the space of Disse is negligible (accounting for
less than 10% of the total liver volume). Such hypotheses, however, do
not hold in the case of liver nodules or during chronic liver diseases,
which modify the sinusoid permeability and the interstitial volume.
Therefore, more complicated models are needed.65,66
Finally, since the liver is approximately 15 cm high, it is not possi-
ble to cover the entire organ even with the larger CT detectors during
a single sequential acquisition. Two strategies can be developed to
elude this limitation: the patient can be moved during the dynamic
acquisition, or two or more consecutive dynamic acquisitions can be
performed. These strategies, however, have to be validated.
CONCLUSION
Functional CT of liver perfusion is a sensitive and specific technique,

which can yield objective and quantitative physiological parameters
and allow repeated longitudinal studies. However, continued improve-
ments and more validation studies are needed to be able to use it on a
routine basis.
Many cancer patients undergo CT imaging surveillance of the liver
for early identification of tumor recurrence following primary treatment.
For colorectal cancer, intensifying follow-up in this way is associated
with reduced mortality,67 and the American Society of Clinical Oncology
now recommends annual CT of the chest and abdomen for 3 years after
primary therapy in patients at higher risk of recurrence.34 CT perfu-
sion could be readily incorporated into such surveillance programs.
The ability to identify a high risk of metastatic liver involvement in a
9781842143094-Ch11 8/8/07 4:25 PM Page 193
frequent disease such as colorectal cancer may help in the decision

whether to use adjuvant chemotherapy, and may avoid unnecessary
treatments in patients who are at low risk of liver metastases.
REFERENCES
1. Paley MR, Ros P. Hepatic metastases. Radiol Clin North Am 1998; 36: 349–363.
2. Baker ME, Pelley R. Hepatic metastases: basic principles and implications for
radiologists. Radiology 1995; 197: 329–37.
3. Finlay IG, McArdle CS. Occult metastases in colorectal carcinoma. Br J Surg
1986; 73: 732–5.
4. Kinkel K, Lu Y, Both M, Warren RS, Thoeni RF. Detection of hepatic metas-
tases from cancers of the gastrointestinal tract by using noninvasive imaging
methods (US, CT, MR imaging, PET): a meta-analysis. Radiology 2002; 224:
748–56.
5. Greenway B. Hepatic metastases from colorectal cancer: resection or not.
Br J Surg 1988; 75: 513–19.
6. Moertel CG. Drug therapy: chemotherapy for colorectal cancer. N Engl J Med
1994; 330: 1136–42.
7. Cooke DA, Parkin A, Wiggins P, Robinson PJ, Giles GR. Hepatic perfusion
index and the evolution of liver metastases. Nucl Med Commun 1987; 8: 970–4.
8. Nott DM, Grime JS, Yates J et al. Changes in the hepatic perfusion index during
the growth and development of experimental hepatic micrometastases. Nucl
Med Commun 1987; 8: 995–1000.
9. Perkins AC, Whalley DR, Ballantyne KC, Hardcastle JD. Reliability of the
hepatic perfusion index for the detection of liver metastases. Nucl Med
Commun 1987; 8: 982–9.
10. Leen E, Goldberg JA, Robertson J et al. Image-directed Doppler ultrasonogra-
phy: a novel technique for the diagnosis of colorectal liver metastases. J Clin
Ultrasound 1993; 21: 221–30.
11. Platt JF, Francis IR, Ellis JH, Reige KA. Liver metastases: early detection based
on abnormal contrast material enhancement at dual-phased helical CT.
Radiology 1997; 205: 49–53.
12. Thimgan MS, Yee HF Jr. Quantitation of rat hepatic stellate cell contraction:
stellate cells, contribution to sinusoidal resistance. Am J Physiol Gastrointest
Liver Physiol 1999; 277: 137–43.
13. Shah V, Haddad FG, Garcia-Cardena G, et al. Liver sinusoidal endothelial cells
are responsible for nitric oxide modulation of resistance in the hepatic sinu-
soids. J Clin Invest 1997; 100: 2923–30.
14. Lautt WW, Legare DJ, Almeida MS. Adenosine as putative regulator of hepatic
arterial flow (the buffer reponse). Am J Physiol 1985; 248: 331–8.
15. Lautt WW. The 1995 Ciba-Geigy Award Lecture. Intrinsic regulation of hepatic
blood flow. Can J Physiol Pharmacol 1996; 74: 223–33.
16. Pandharipande PV, Krinsky GA, Rusinek H, Lee VS. Perfusion imaging of the
liver: current challenges and future goals. Radiology. 2005; 234: 661–73.
obtained with dynamic CT. Radiology 1993; 188: 405–11.
9781842143094-Ch11 8/8/07 4:25 PM Page 194
18. Dugdale PE, Miles KA. Hepatic metastases: the value of quantitative assess-
ment of contrast enhancement on computed tomography. Eur J Radiol 1999;
30: 206–13.
19. Blomley M, Coulden R, Dawson P. Liver perfusion studied with ultrafast CT.
20. Tsushima Y, Blomley JK, Kusano S, Endo K. The portal component of hepatic
perfusion measured by dynamic CT: an indicator of hepatic parenchymal
damage. Dig Dis Sci 1999; 44: 1632–8.
21. Miles KA, McPherson SJ, Hayball MP. Transient splenic inhomogeneity on
contrast enhanced CT: mechanism and effect of liver disease. Radiology 1995;
194: 91–5.
22. Leggett DA, Kelley BB, Bunce IH, Miles KA. Colorectal cancer: diagnostic
potential of CT measurements of hepatic perfusion and implications for con-
trast enhancement protocols. Radiology 1997; 205: 716–20.
23. Miles KA, Leggett DA, Kelley BB et al. In vivo assessment of neovasculariza-
tion of liver metastases using perfusion CT. Br J Radiol 1998; 71: 276–81.
24. Halavaara JT, Hamberg LM, Leong FS et al. Functional CT with an experimen-
tal intravascular contrast agent in the assessment of liver vascular physiology.
Acad Radiol 1996; 3: 946–52.
25. Materne R, Van Beers BE, Smith AM et al. Non-invasive quantification of
liver perfusion with dynamic computed tomography and a dual-input one-
compartmental model. Clin Sci (Lond) 2000; 99: 517–25.
26. Van Beers BE, Leconte I, Materne R et al. Hepatic perfusion parameters
in chronic liver disease: dynamic CT measurements correlated with disease
severity. AJR Am J Roentgenol 2001; 176: 667–73.
27. Cuenod C, Leconte I, Siauve N et al. Early changes in liver perfusion caused
by occult metastases in rats: detection with quantitative CT. Radiology 2001;
218: 556–561.
28. Cuenod CA, Leconte I, Siauve N et al. Deconvolution technique for measuring
tissue perfusion by dynamic CT: application to normal and metastatic liver.
Acad Radiol 2002; 9: S205–11.
29. Fournier LS, Cuenod CA, de Bazelaire C et al. Early modifications of hepatic
perfusion measured by functional CT in a rat model of hepatocellular carci-
noma using a blood pool contrast agent. Eur Radiol 2004; 14: 2125–33.
30. Axel L. Tissue mean transit time from dynamic computed tomography by a
simple deconvolution technique. Invest Radiol 1983; 18: 94–19.
31. Zierler KL. Equation for measuring blood flow by external monitoring of
radioisotopes. Circ Res 1965; 16: 309–21.
32. Miles KA, Kelley BB. Altered perfusion adjacent to hepatic metastases. Clin
Radiol 1997; 52: 162–3.
33. Tsushima Y, Blomley MJ, Yokoyama H, Kusano S, Endo K. Does the presence
of distant and local malignancy alter parenchymal perfusion in apparently
disease-free areas of the liver? Dig Dis Sci 2001; 46: 2113–19.
34. Desch CE, Benson AB 3rd, Somerfield MR et al. American Society of Clinical
Oncology. Colorectal cancer surveillance: 2005 update of an American Society
of Clinical Oncology practice guideline. J Clin Oncol, 2005; 23: 8512–19.
35. Compton CC, Fielding LP, Burgart LJ et al. Prognostic factors in colorectal
cancer. College of American Pathologists Consensus Statement 1999. Arch
Pathol Lab Med 2000; 124: 979–94.
9781842143094-Ch11 8/8/07 4:25 PM Page 195
36. Garcea G, Sharma RA, Dennison A et al. Molecular biomarkers of colorectal

carcinogenesis and their role in surveillance and early intervention. Eur J
Cancer 2003; 39: 1041–52.
37. Smith JJ, Sorensen AG, Thrall JH. Biomarkers in imaging: realizing radiology’s
future. Radiology 2003; 227: 633–8.
38. Mooteri S, Rubin D, Leurgans S et al. Tumor angiogenesis in primary and
metastatic colorectal cancers. Dis Colon Rectum 1996; 39: 1073–80.
39. Miles KA, Colyvas K, Griffiths MR, Bunce IH. Colon cancer: Risk stratification
using hepatic perfusion CT. European Radiology 2004; 14: Suppl 2, 129.
40. Fuentes MA, Keith CJ, Griffiths M, Durbridge G, Miles KA. Hepatic haemody-
namics: interrelationships between contrast enhancement and perfusion on CT
and Doppler perfusion indices. Br J Radiol 2002; 75: 17–23.
41. Ballantyne KC, Charnley RM, Perkins AC et al. Hepatic perfusion index in
the diagnosis of overt metastatic colorectal cancer. Nucl Med Commun 1990;
11: 23–8.
42. Hugier M, Maheswari S, Toussaint P et al. Hepatic flow scintigraphy in evalu-
ation of hepatic metastases with gastrointestinal malignancy. Arch Surg 1993;
128: 1057–9.
43. Carter R, Hemingway D, Cooke TG et al. A prospective study of six methods
for detection of hepatic colorectal metastases. Ann R Coll Surg Engl 1996; 78:
27–30.
44. Leen E, Goldberg JA, Anderson JR et al. Hepatic perfusion changes in patients
with liver metastases: comparison with those patients with cirrhosis. Gut 1993;
34: 554–7.
45. Kissel A, Rixe O, Methlin A et al. Quantification of hepatic arterial and portal
venous flow using ultrasound contrast agents for early detection of liver
metastases of colorectal cancers. J Radiol 2001; 82: 1621–5.
46. Leen E, Goldberg JA, Angerson WJ, McArdle CS. Potential role of Doppler
perfusion index in selection of patients with colorectal cancer for adjuvant
chemotherapy. Lancet 2000; 355: 34–37.
47. Roumen RM, Scheltinga MR, Slooter GD, van der Linden AW. Doppler perfu-
sion index fails to predict the presence of occult hepatic colorectal metastases.
Eur J Surg Oncol 2005; 31: 521–7.
48. Yarmenitis SD, Kalogeropoulou CP, Hatjikondi O et al. An experimental
approach of the Doppler perfusion index of the liver in detecting occult
hepatic metastases: histological findings related to the hemodynamic measure-
ments in Wistar rats. Eur Radiol 2000; 10: 417–24.
49. Totman JJ, O’Gorman RL, Kane PA, Karani JB. Comparison of the hepatic per-
fusion index measured with gadolinium-enhanced volumetric MRI in controls
and in patients with colorectal cancer. Br J Radiol 2005; 78: 105–9.
50. Breedis C, Young G. The blood supply of neoplasms in the liver. Am J Pathol
1954; 30: 969–77.
51. Fong Y. Doppler perfusion index in colorectal cancer. Lancet 2000; 355: 5–6.
52. Warren HW, Anderson WJ, Leen E et al. Haemodynamic changes associated
with colorectal liver metastases: evidence of a humoral mediator. Br J Surg
1993; 80: 1561(abstr).
53. Carter R, Anderson JH, Cooke TG, Baxter JN, Angerson WJ. Splanchnic blood
flow changes in the presence of hepatic tumour: evidence of a humoral
mediator. Br J Cancer 1994; 69: 1025–6.
9781842143094-Ch11 8/8/07 4:25 PM Page 196
54. Ridge JA, Bading JR, Gelbard AS, Benua RS, Daly J. Perfusion of colorectal
hepatic metastases: relative distribution from the hepatic artery and portal vein.
Cancer 1987; 59: 1547–53.
55. Kan Z, Ivancev K, Lunderquist A et al. In vivo microscopy of hepatic tumors
in animal model: a dynamic investigation of blood supply to hepatic
metastases. Radiology 1993; 187: 621–6.
56. Morris VL, MacDonald IC, Koop S et al. Early interaction of cancer cells with
the microvasculature in mouse liver and muscle during hematogenous metas-
tasis: videomicroscopic analysis. Clin Exp Metastasis 1993; 11: 377–50.
57. Kruskal JB, Thomas P, Kane RA, Goldberg SN. Hepatic perfusion changes in
mice livers with developing colorectal cancer metastases. Radiology 2004; 231:
482–90.
58. Kopljar M, Brkljacic B, Doko M, Horzic M. Nature of Doppler perfusion index
changes in patients with colorectal cancer liver metastases. J Ultrasound Med
2004; 23: 1295–300.
59. Tian JL, Zhang JS. Hepatic perfusion disorders: etiopathogenesis and related
diseases. World J Gastroenterol 2006 May 28; 12: 3265–70.
60. Haugeberg G, Strohmeyer T, Lierse W. The vascularization of liver metastases.
J Cancer Res Clin Oncol 1988; 114: 415–19.
61. Terayama N, Terada T, Nakanuma Y. A morphometric and immunohisto-
chemical study on angiogenesis of human metastatic carcinomas of the liver.
Hepatology 1996; 24: 816–19.
62. Takeda A, Stoeltzing O, Ahmad SA et al. Role of angiogenesis in the develop-
ment and growth of liver metastasis. Ann Surg Oncol 2002; 9: 610–16.
63. Bader TR, Grabenwoger F, Prokesch RW, Krause W. Measurement of hepatic
perfusion with dynamic computed tomography: assessment of normal values
and comparison of two methods to compensate for motion artifacts. Invest
Radiol 2000; 35: 539–47.
64. Nakashige A, Horiguchi J, Tamura A et al. Quantitative measurement of
hepatic portal perfusion by multidetector row CT with compensation for
respiratory misregistration. Br J Radiol 2004; 77: 728–34.
65. Kapanen MK, Halavaara JT, Hakkinen AM. Assessment of vascular physiology
of tumorous livers: comparison of two different methods. Acad Radiol 2003;
10: 1021–9.
66. Kapanen MK, Halavaara JT, Hakkinen AM. Open four-compartment model in
the measurement of liver perfusion. Acad Radiol 2005; 12: 1542–50. Erratum
in: Acad Radiol 2006; 13: 270.
67. Jeffrey GM, Hickey BE, Hider P. Follow-up strategies for patients treated for
non-metastatic colorectal cancer (Cochrane Review). In: The Cochrane Library,
Issue 2, 2003. Oxford: Update Software, 2003.
9781842143094-Ch12 8/8/07 12:07 PM Page 197
12
Beyond RECIST: CT perfusion
in evaluating treatment
response and complications
Natalie Charnley and Kenneth A Miles
INTRODUCTION
Response assessment in oncology, using conventional morphological

imaging, relies on tumor shrinkage. This assessment has been standard-
ized using the Response Evaluation Criteria in Solid Tumors (RECIST),
which have recently been updated to reflect more accurately the change
in volume of spherical lesions.1 Despite this change, RECIST measurements
are prone to errors, are particularly difficult for small lesions, and are
laborious. Furthermore malignant lesions may be difficult to distinguish
from areas of fibrosis.1 Functional imaging using perfusion computed
tomography (CTP), dynamic contrast-enhanced magnetic resonance
imaging (DCE-MRI), [15O]H2O-positron emission tomography (PET), or
ultrasound (US) may provide more specific information about changes in
tumor biology. Tumor shrinkage may be a valid endpoint for established
cytotoxic drugs, but may not be relevant for biological agents under devel-
opment, which are often cytostatic. In addition, functional imaging may
be more sensitive in detecting an early response to treatment and can
potentially identify subpopulations of patients enriched for response.
Perfusion describes the flow of blood through tissues, and is gov-
erned by the tumor vasculature. Tumor blood vessels have structural
and functional abnormalities including irregular branching, increased
permeability, and independence from normal flow control mechanisms.
This leads to variable and inadequate delivery of nutrients to the tumor
tissue. Increased permeability of tumor vasculature with respect to
normal tissue causes differential uptake of contrast agents, which is
seen on functional imaging.
9781842143094-Ch12 8/8/07 12:07 PM Page 198
The importance of tumor vasculature in supporting growth and the

associated therapeutic implications have long been recognized.2 Several
agents, including conventional cytotoxic drugs, can effect changes in
tumor vasculature.3,4 Over the past decade, there has been increasing
interest in the development of specific antiangiogenic drugs and vascu-
lar disrupting agents (VDAs), which impair angiogenesis and destroy
existing vasculature, respectively. There is an increasing need for direct
imaging of the biological endpoints of these drugs, in particular tumor
perfusion and endothelial permeability. Furthermore, vascular changes
have been observed in tumors following radiotherapy.5,6 There is inter-
est in monitoring these changes to predict radiation response and also
in assessing perfusion of normal tissue in the evaluation of radiotherapy-
induced complications.
Computed tomography (CT) remains the mainstay technique for
response evaluation in oncology, and is widely used for RECIST meas-
urements. CT perfusion can be readily included as part of a conven-
tional CT examination to provide a widely available and relatively cheap
means of obtaining a simultaneous functional response assessment.
Interestingly, there are no data available for the in vivo reproducibility
of manual RECIST measurements in which the same subject has under-
gone CT twice (test–retest protocol). Yet it is well recognized that differ-
ences in slice positioning can produce apparent changes in tumor size.
Even repeated size measurements from the same tumor study show
considerable variability.7 The test–retest performance of CT perfusion is
discussed more fully in Chapter 2. Nevertheless, the coefficient of
repeatability expressing the 95% confidence limits of individual CT per-
fusion measurements in humans has been reported to be 28% for normal
brain (Griffiths, personal communication) and 65% for rectal cancer.8
CTP is especially well suited to assessing the response to agents which
act on tumor vasculature. CTP can provide a range of parameters relevant
to tumor perfusion, namely blood flow (BF), blood volume (BV), mean
transit time (MTT), and permeability surface area (PS), and can provide
quantitative data. Commercial analysis software approved by the US Food
and Drug Administration (FDA) is now available, which makes data
analysis more feasible. Importantly, studies have shown that CTP-derived
parameters correlate with histological measurements of angiogenesis.9,10,40
CT PERFUSION IN ASSESSING TUMOR RESPONSE TO DRUGS

WHICH AFFECT TUMOR VASCULATURE
Some of the earliest studies of CTP in oncology investigated agents

which affect the blood–brain barrier (BBB) in patients with brain tumors.
9781842143094-Ch12 8/8/07 12:07 PM Page 199
Beyond RECIST: evaluating treatment response 199
In one of these studies, perfusion parameters were investigated in 10

patients with primary brain tumors after a week of oral dexamethasone.3
Both the transfer constant and the plasma volume within the tumors
decreased with steroid treatment. Interestingly, when enhancement was
assessed on contrast-enhanced CT scans, no change in enhancement
was visible in the majority of cases.
A further study of CTP in patients with brain tumors assessed BBB
permeability following administration of the agent RMP-7, a bradykinin
analog.11 Chemotherapeutic agents traditionally have poor penetration
of the BBB, and so a drug which can break down the tumor BBB
would potentially improve delivery of the cytotoxic agent. CTP demon-
strated an increase in BBB permeability with RMP-7 (Figure 12.1). On
the background of this finding, RMP-7 has been taken forward into
further clinical development with the cytotoxic carboplatin, and is now
in phase II trials.12
Tumor hypoxia adversely affects prognosis, and is associated with
reduced survival following surgery or radiotherapy.13,14 Drugs are avail-
able which can preferentially target and destroy hypoxic cells. BW12C
is an agent which binds preferentially to oxyhemoglobin and causes
reduced tissue oxygenation.15 Thirty-two patients with advanced gastro-
intestinal cancer were recruited to a phase I trial of BW12C plus a
standard cytotoxic, mitomycin C, to assess effects of the drug on
oxygenation and tumor perfusion.15 Infusion of BW12C was followed
after 2 hours by imaging with CTP. In five out of six patients who were
eligible for the imaging study perfusion was reduced by 30%, though
a 140 b 14
120 12
Permeability (µl/min/ml)
Permeability (µl/min/ml)
100 10
80 8
60 6
40 4
20 2
0 0
Baseline With RMP-7 Baseline With RMP-7
Figure 12.1 Changes in the vascular permeability of gliomas (a) and normal
brain tissue (b) following RMP-7 assessed using computed tomography perfusion
(CT). (Data from reference 11)
9781842143094-Ch12 8/8/07 12:07 PM Page 200
1.5
ml/min/ml
0.5
Figure 12.2 Changes in tumor perfusion following BW12C assessed by

CT perfusion. (Reproduced with permission from reference 16)
this change was not significant. However, the greatest reductions in

perfusion were seen in those tumors exhibiting higher baseline values
(Figure 12.2).
There have been several studies where functional imaging has been
used to assess vascular response to conventional cytotoxic agents.
Many of these studies have been in the neoadjuvant treatment of breast
cancer, where the rationale is to establish a biomarker of response to
standard chemotherapy prior to surgery.4,17 A single study has used CTP
for this purpose.18 Dynamic CT was performed in 19 patients prior
to and following standard cytotoxic treatment with anthracycline and
taxane, and contrast enhancement was compared with residual tumor
extent postoperatively. Perfusion parameters were not actually derived
in this study, but the pattern of contrast enhancement at early and late
phases was assessed. For late-phase CTP, the accuracy of detecting
residual disease was 100% (7/7) where there was diffuse contrast
enhancement in a whole quadrant prior to chemotherapy, and 82%
(9/11) where contrast enhancement was localized.
The ability of CTP to assess the response of squamous cell carcino-
mas of the oropharynx to induction chemotherapy has been recently
assessed by Gandhi et al.19 CTP was performed in a series of nine
patients of stage 3 or 4, before and 3 weeks after one cycle of induc-
tion chemotherapy. A reduction in BV of 20% or more showed a sub-
stantial agreement with clinical response assessed by endoscopy. Two
small studies have also highlighted the potential for CTP to evaluate
9781842143094-Ch12 8/8/07 12:07 PM Page 201
Figure 12.3 CT perfusion images of non-small-cell lung cancer before (a) and
after (b) chemotherapy
the treatment response in lung cancer (Figure 12.3). Choi et al. meas-
ured peak tumor enhancement in residual masses in seven patients with
small-cell cancer following chemotherapy.20 Peak enhancement values
increased in four, decreased in one, and were unchanged in two. All
four patients demonstrating an increase in enhancement were reported
to exhibit improvement. As part of a larger study, Kiessling et al. pre-
sented a single case of non-small-cell lung cancer in which a partial
response on size criteria was associated with a reduction in perfusion.21
CT PERFUSION IN ASSESSING TUMOR RESPONSE TO NOVEL

ANTIVASCULAR AGENTS IN PHASE I TRIALS
Standard clinical development of novel drugs begins in phase I trials.

The main purpose of these is to ascertain the range of toxic effects of
that drug, and its maximum tolerated dose (MTD), thought to be the
most effective dose. The pharmacokinetics and pharmakodynamics of
9781842143094-Ch12 8/8/07 12:07 PM Page 202
that agent are also assessed, as are early response data. MTD is relevant
to treatment with cytotoxic drugs, but not to many biological agents
such as antiangiogenic drugs and VDAs. Furthermore, there has been
no correlation between the activity of antiangiogenic agents and their
toxicity.22 Instead, surrogate endpoints must be investigated.
Angiogenesis is the formation of new tumor blood vessels, essential
for tumor growth from 1 mm3 in size.2 The trigger for angiogenesis
is additionally determined by hypoxia, acidosis, and hypoglycemia.23
The process of angiogenesis is tightly coordinated by a network of
cytokines. Vascular endothelial growth factor (VEGF) stimulates angio-
genesis and also has effects on vascular permeability. Fibroblast growth
factor (FGF) and platelet-derived growth factor (PDGF) also have a role
in angiogenesis, but are not vascular endothelium-specific. Integrins
such as αvβ3, which is expressed selectively on activated tumor vascu-
lature, potentiate angiogenesis and enable endothelial cell migration
towards the tumor.
Many of the steps in the angiogenic pathway have been targeted by
novel antiangiogenic drugs, and several of these are in clinical devel-
opment.22,24–27 Several of these trials compare CTP to other functional
imaging such as DCE-MRI, and also to histology, in proof of principle
studies. A summary of these trials is given in Table 12.1.
Willett’s group25 conducted a phase I trial of of bevacizumab, a VEGF-
specific antibody, in six patients with rectal adenocarcinoma. Patients
received a regimen of preoperative bevacizumab and chemoradiotherapy.
Perfusion was assessed initially 12 days following a single infusion of
the antiangiogenic agent. CTP revealed a significant decrease in tumor
perfusion of 40–44% and of blood volume of 16–39%, in four out of five
patients assessed. This was associated with a significant decrease in
tumor microvessel density (MVD). In addition, a fall in tumor intersti-
tial fluid pressure, and an increase in vessels positive for smooth muscle
actin, were observed. The authors suggest that this is supportive of
vascular normalization with bevacizumab.
Meijerink et al.24 assessed the perfusion of a range of advanced solid
tumors in a phase I trial of 13 patients using a combination of agents:
gefitinib, a tyrosine kinase inhibitor of epidermal growth factor receptor
1 (EGFR1), and the antiangiogenic agent AZD21271, a tyrosine kinase
inhibitor of VEGFR1–3. Both agents were given daily. CTP was per-
formed prior to starting treatment, and 4–6 weeks following treatment.
A longer scanning protocol was used for patients with liver metastases
compared to other sites, in order to determine the arterial and portal
fractions of tumor blood flow separately. In five out of six patients with
extrahepatic masses, there was an average decrease in perfusion of
9781842143094-Ch12
8/8/07
Table 12.1 Monitoring drug treatment using perfusion CT (CTP)
Study n Treatment Tumor site Agent Follow-up CTP findings
Yeung, 19943 10 Dexamethasone Brain Steroid 1 week ↓Transfer constant, ↓plasma volume
12:07 PM
Ford, 199611 4 RMP-7 Brain Bradykinin analog 1h ≠BBB permeability

Meijerink, 200624 13 Gefitinib + AZD2171 Various AA 4–6 weeks ↓BF in 5/6 with extrahepatic mets,
reversible ↓perfusion in hepatic mets
Thomas, 200327 21 Endostatin Various AA 8 weeks ↓BF in 4 patients
Page 203
Willett, 200425 5 Bevacizumab Rectum AA 12 days 4/5 ↓tumor BF, BV, MVD
McNeel, 200526 25 MEDI-522 Various AA 8 weeks ≠MTT with ≠doses of drug
Xiong, 200422 6 SU6668 Various AA 4 weeks, 5/6 ↓flow on CT
12 weeks
Falk, 199415 6 BW12C GI Binds to OHb 1h NS ↓tumor BF in 5/6 patients
Tozaki, 200418 19 Anthracycline + Breast Neoadjuvant NSt 82–100% accurate in detecting residual
taxane cytotoxic disease postoperatively
GI, gastrointestinal; AA, antiangiogenic agent; OHb, oxyhemoglobin; NSt, not stated: assessment made after 4–6 cycles of chemotherapy, likely to be 12–18 weeks;
BBB, blood–brain barrier; BF, blood flow; mets, metastases; BV, blood volume; MVD, microvessel density; MTT, mean transit time; NS, non-significant
Beyond RECIST: evaluating treatment response
203
9781842143094-Ch12 8/8/07 12:07 PM Page 204
18% followed by a plateau. In liver metastases, an initial average

decrease in perfusion of 39% was observed followed by a return to
previous hepatic artery blood flow. Importantly, no correlation was
seen between changes in tumor size as assessed by RECIST and change
in perfusion.
MEDI-522, a monoclonal antibody to the αvβ3 integrin expressed
selectively on tumor neovasculature, was evaluated in a phase I trial of
25 patients with a variety of solid tumors.26 Patients underwent CTP at
baseline and following 8 weeks of weekly MEDI-522 given as a range
of doses. Tumor blood flow, blood volume, mean transit time, and per-
meability surface area were also assessed. Only 11 patients had tumors
which satisfied the size criteria for follow-up CT scanning. A positive
relationship was seen between increasing dose of drug and mean tran-
sit time, but not the other parameters. This would suggest a MEDI-522-
related reduction in blood flow through small tumor vessels. No
complete or partial responses were observed by RECIST criteria, but three
patients experienced prolonged stable disease. There was no correlation
between CT parameters and MEDI-522 serum concentration.
Endostatin is an endogenous antiangiogenesis agent. A phase I clin-
ical trial of recombinant human endostatin was conducted by Thomas’s
group27 in 21 patients with advanced solid tumors. Patients received
the drug daily for a 28-day cycle, and at 8 weeks underwent imaging
with CTP, DCE-MRI, fluorodeoxyglucose (FDG)-PET, and US. In four
patients, a reduction in tumor perfusion was seen on the CT time–atten-
uation curve, which was suggestive of a decrease in microvessel density.
No changes were seen on MRI or US at this time, and there was no
change in MVD to corroborate the CT findings. In addition, the change
in the CTP time–attenuation curves did not predict response to treatment.
Xiong et al.22 investigated the pharmacokinetics and pharmacody-
namics of SU6668, a novel tyrosine kinase inhibitor of VEGFR, fibro-
blast growth factor receptor (FGFR), and platelet-derived growth factor
receptor (PDGFR). The drug was given daily in a 4-week cycle, and
tumor perfusion was assessed by CTP and DCE-MRI at 4 and 12 weeks.
Five out of six evaluable patients had a reduction in tumor blood flow
on CTP at 4 and 12 weeks. Two out of four patients showed an
equivalent response on MRI. Changes in functional imaging were
not matched with any clinical responses in this study.
Unfortunately, several of these trials have had small numbers of
patients, and sometimes have not had the power to detect significant
changes in perfusion parameters. It is difficult to make comparisons
between results of phase I trials. Different perfusion parameters are
often chosen for these studies, and the specific CTP protocol and data
9781842143094-Ch12 8/8/07 12:07 PM Page 205
model used are not always described. However, none of the above
studies in which multiple parameters were studied showed significant
changes in vascular permeability on treatment, even when significant
changes in other parameters had been observed.22,25,26 This finding is
interesting, considering the importance of VEGF not only in stimulating
angiogenesis but also in determining capillary leakiness. CT measure-
ments of permeability effectively describe the rate of transfer of contrast
material between intravascular and extravascular compartments. It is
feasible that in some circumstances, a decrease in transport of contrast
material due to a treatment-induced reduction in vascular permeability
may be offset by an increase in transport resulting from a concomitant
reduction in tumor interstitial fluid pressure.
When evaluating tumor response, the time scale for imaging is of
paramount importance. Frequently, changes are assessed at times when
we would expect to see a response on conventional cross-sectional
imaging, but this may not be the most appropriate time point for anti-
angiogenic drugs and VDAs. Response should not be assessed too
early, as during the first few days following an antiangiogenic drug, it
is thought that the vasculature undergoes some normalization,28 which
may lead to a reduction in interstitial fluid pressure and an initial pos-
sible increase in perfusion. However, assessing at late time points may
cause a change in tumor size and difficulty in co-registering the regions
of interest.
CT PERFUSION IN ASSESSING TUMOR RESPONSE TO

RADIOTHERAPY
Radiotherapy has the potential to cause changes in tumor and normal

tissue vasculature by its ability to cause breaks in DNA. The ability of
radiation to damage blood vessels has long been recognized, as evi-
denced by the late effect of radiation-induced telangiectasia. In the case
of vessel damage, it would be expected that tumor and normal tissue
blood flow would decrease. However, paradoxically, an early effect
of radiation could be an increase in blood flow, brought about by
radiotherapy-induced inflammation, and also a decrease in tumor inter-
stitial pressure which may open compressed tumor vessels and allow
reperfusion.
Three studies have used CTP to monitor the effects of radiotherapy
on tumor vasculature and response to treatment. As with monitoring drug
response, the direction of change in perfusion parameters is affected
by the timing of assessment. Harvey’s group investigated tumor blood
9781842143094-Ch12 8/8/07 12:07 PM Page 206
flow in 22 patients with carcinoma of the prostate, both 1–2 weeks and
6–12 weeks following completion of treatment.29 Increases in perfusion,
contrast agent clearance, and fractional vascular volume were observed
1–2 weeks following treatment, and these were maintained at the later
time point. Increases in contrast agent clearance and fractional vascular
volume, but not perfusion, were also seen at the same time points in
other tumor types.30
CTP has also been used to assess patients undergoing combined
modality treatment with chemoradiotherapy.31 Nine patients with rectal
carcinoma underwent a 6–8-week course of chemoradiation prior to
surgery. Patients were assessed following completion of treatment.
A reduction in BF and increase in MTT were observed, suggesting
damage to vasculature. BV and PS remained constant.
Response to other forms of radiation treatment can also be studied
using CTP. Bondestam et al.32 have investigated perfusion parameters
in six patients with meningiomas, undergoing brachytherapy treatment
by stereotactic implantation of iodine-125 seeds. A 41% reduction in
tumor blood flow was seen 3 months following the procedure using
CTP imaging.
OTHER THERAPIES
Other tumor treatment modalities evaluated by CT perfusion include

transcatheter arterial chemoembolization and laser thermal therapy.
Tsushima et al. used CT perfusion to study the changes in hepatic
perfusion following transcatheter arterial chemoembolization of liver
tumors using adriamycin and lipiodol.33 Mean arterial perfusion in four
tumors fell from 90 ml/min/100 ml to 28 ml/min/100 ml. The potential
utility of CT perfusion for monitoring laser thermal therapy has been
demonstrated in animals.34 The mean decrease in tumor perfusion was
41 ± 14%. An even greater reduction in tumor perfusion (64 ± 10%,
p < 0.001) was produced by hypocapnia, which minimized heat dissipa-
tion through blood flow, resulting in a significantly larger thermal lesion.
IDENTIFICATION OF SUBPOPULATIONS ENRICHED

FOR RESPONSE
The relationship between tumor vascularity and treatment response is

complex. Highly vascularized tumors tend to be more aggressive, and
therefore potentially less likely to respond to treatment. However, a
9781842143094-Ch12 8/8/07 12:07 PM Page 207
well-developed tumor blood supply is also essential for the effective

delivery of chemotherapeutic drugs to tumor tissue, and thus poorly
perfused tumors may also be considered less likely to respond.
Furthermore, low levels of perfusion and/or a heterogeneous vascular
network are frequently associated with areas of tumor hypoxia, a fea-
ture that is known to confer resistance to not only chemotherapy but
also radiotherapy. Thus, CT perfusion measurements have the potential
to predict tumor response to chemotherapy or radiotherapy with low
and/or heterogeneous contrast enhancement and perfusion, implying a
likelihood of an unfavorable outcome.
Three preliminary studies illustrate the potential for CT perfusion to
identify patients less likely to respond to chemotherapy. In a cohort of
patients with small-cell lung cancer, Choi et al. determined peak con-
trast enhancement from spiral CT acquisitions performed at 40 s and
2–3 min following a 120-ml bolus of contrast medium (300 mg/ml).20
Peak tumor enhancement correlated with the subsequent reduction in
tumor volume following chemotherapy (r = 0.57, p < 0.002). Ten of 11
(91%) of patients with tumor enhancement less than 30HU failed
to achieve a reduction in tumor volume of 70% or more (Figure 12.4).
A small study of 22 patients has shown the potential of CT measure-
ments of hepatic arterial and portal perfusion to predict chemo-
therapeutic response in colorectal cancer.35 Portal perfusion was
significantly lower in those patients whose disease progressed despite
treatment (29 versus 42 ml/min/100 ml, p < 0.05). A portal perfusion
value of 30 ml/min/100 ml had a predictive value for progression
despite treatment of 80%. More recently, in a study of nine patients with
stage 3 or 4 squamous cell carcinomas of the oropharynx, Gandhi et al.
showed that higher tumor blood volume measurements on CTP were
associated with a greater endoscopic tumor response (p < 0.05) follow-
ing induction chemotherapy.19
The ability of CT perfusion values to predict response to radiother-
apy has been shown by Hermans et al. in a study of 105 patients with
head and neck cancer undergoing radiotherapy with curative intent.36
In multivariate analysis, low perfusion (along with T-stage) was an
independent predictor of local control but not of regional control or
cause-specific survival. Within T-stages 3 and 4, perfusion values could
stratify patients according to likely response. Amongst 34 patients with
T-4 tumors, the likelihood of local control at 18 months was less than
10% if pre-treatment tumor perfusion was less than the median value of
83.5 ml/min/100 g.
Sahani et al. related the baseline CT perfusion values in rectal can-
cers to the subsequent response to chemoradiation.31 However, the
9781842143094-Ch12 8/8/07 12:07 PM Page 208
a b 50HU
0HU
d 50HU
c
0HU
Figure 12.4 Conventional CT (a, c) and maximum enhancement images

(b, d) of two cases of small-cell lung cancer. The tumor in the upper row shows
higher enhancement than the tumor in the lower row and is therefore more
likely to respond to chemotherapy. (Maximum enhancement images displayed
with identical windows)
results are in contrast with the studies above in that a high initial BF
and a short MTT were associated with poor response. The presence of
intratumoral arteriovenous shunts was proposed as a possible explana-
tion for these findings.
The use of imaging to identify patients unlikely to respond to cancer
therapy has the potential to save the morbidity and cost of futile treat-
ment. However, the diagnostic thresholds chosen must favor high
specificity over sensitivity in order to ensure that treatment is not with-
held from potential responders in error. This use of imaging is also
affected by the prevalence of non-responders in the treated population,
and is more useful when the proportion of patients failing to respond
9781842143094-Ch12 8/8/07 12:07 PM Page 209
is moderate or high. By reflecting the vascular heterogeneity that leads

to hypoxia, the future development of parameters that simultaneously
encapsulate both the intensity and heterogeneity of contrast enhance-
ment and/or perfusion may be able to improve these predictive values
further.
ASSESSMENT OF TREATMENT COMPLICATIONS
The toxic effects of chemotherapeutic agents and radiotherapy may be

associated with changes in organ perfusion. Such changes are poten-
tially assessable with CT perfusion, either as part of a tumor evaluation
study or as a primary study of treatment toxicity. Millar et al. correlated
CTP parameters with symptoms of radiation toxicity in 14 patients
with cerebral metastases undergoing short-course palliative cranial
irradiation.37 A 19% increase was observed in PS at day 2, which
returned to its original value on day 5. MTT was negatively correlated
with headache score, whereas nausea tended to be associated with
increased blood volume. The high spatial resolution afforded by CTP
enables assessment of the changes in intrarenal perfusion to be associ-
ated with drug nephrotoxicity.38 Milder degrees of nephrotoxicity may
be associated with greater reductions in cortical than in medullary
perfusion (Figure 12.5).
Bone-marrow transplantation is increasingly used in the management
of hematological malignancies. Veno-occlusive disease and graft-versus-
host disease are important hepatic complications of this procedure. The
measurement of liver hemodynamics can contribute to the assessment
of these conditions. Figure 12.6 illustrates the ability of CT perfusion to
depict the changes in hepatic arterial and portal perfusion in a case of
graft-versus-host disease.
CONCLUSIONS
In conclusion, CTP is a versatile and validated technique which can

be used to assess several parameters reflecting changes in tumor vascu-
lature. CTP has the potential to monitor a range of drugs which can
affect tumor vasculature, including antiangiogenic agents with various
mechanisms of action on the angiogenic network. There is also the
potential to evaluate radiotherapy-induced changes in tumor vascula-
ture, and also the perfusion of normal tissue in the assessment of radio-
therapy-induced complications. The use of CT perfusion to identify
9781842143094-Ch12 8/8/07 12:07 PM Page 210
4
Perfusion (ml/min/ml)
0
1 1 2 2 3 4 4 5 Norm
Patient number
Figure 12.5 (a) CT perfusion image in a patient with nephrotoxicity induced

by cyclosporine. The effects of cyclosporine are more readily seen in cortical per-
fusion values (b). Data are arranged in order of decreasing blood levels from
left to right. Norm, normal values. (Reproduced with permission from reference 16)
subpopulations enriched for response offers the prospect of personal-

ized cancer care and improved cost-effectiveness.
To date, the evaluation of tumor response with CT perfusion has mostly
been applied in the early stages of drug development to provide proof
of biological activity. Such evidence of activity can impact significantly
9781842143094-Ch12 8/8/07 12:07 PM Page 211
a b
c d
Figure 12.6 Hepatic arterial (a) and portal (b) perfusion images in a patient
with graft-versus-host disease following bone-marrow transplantation. Hepatic
arterial perfusion is markedly reduced throughout the liver. The changes are
comparable to those seen in liver allograft rejection (c, d)
on whether a particular agent progresses to late-stage trials. The sim-

plicity and wide availability of CT perfusion also makes the technique
ideally suited to tumor response evaluation in late-stage trials and
clinical practice. However, there remains a need for further evidence
linking CT perfusion responses to ultimate outcomes such as survival.
In the future, technology should be developed to enable the imag-
ing of specific angiogenic pathways by linking contrast agent to angio-
genic markers, as has been achieved with DCE-MRI.39 Protocols need
to be standardized so that comparisons can be made between trials (see
Chapter 3), and also efforts should be made to elucidate the optimal
time for imaging these agents.
REFERENCES
1. Padhani A, Ollivier L. The RECIST (Response Evaluation Criteria in Solid Tumors)

criteria: implications for diagnostic radiologists. Br J Radiol 2001; 74: 983–6.
9781842143094-Ch12 8/8/07 12:07 PM Page 212

285: 1182–6.
3. Yeung WT, Lee TY, Del Maestro RF et al. Effects of steroids on iopamidol
blood–brain barrier transfer constant and plasma volume in brain tumors
measured with X-ray computed tomography. J Neurooncol 1994; 18: 53–60.
4. Delille JP, Slantetz PJ, Yeh ED et al. Invasive ductal breast carcinoma response
to neoadjuvant chemotherapy: non invasive monitoring with functional MR
imaging pilot study. Radiology 2003; 228: 63–9.
5. Hoskin PJ, Saunders MI, Goodchild K et al. Dynamic contrast enhanced mag-
netic resonance scanning as a predictor of response to accelerated radiother-
apy for advanced head and neck cancer. Br J Radiol 1999; 72: 1093–8.
6. Mineura K, Yasuda T, Kowada M et al. Positron emission tomographic evalu-
ation of radiochemotherapeutic effect on regional cerebral hemocirculation
and metabolism in patients with glioma. J Neurooncol 1987; 5: 277–85.
7. Erasmus JJ, Gladish GW, Broemeling L et al. Interobserver and intraobserver
variability in measurement of non-small-cell carcinoma lung lesions: implica-
tions for assessment of tumor response. J Clin Oncol 2003; 21: 2574–82.
8. Goh V, Halligan S, Gartner L, Bassett P, Bartram Cl. Quantitative colorectal
cancer perfusion management by multidetector-row CT: does greater tumor
9. Hayashi K, Tozaki M, Sugisaki M et al. Dynamic multislice helical CT of
ameloblastoma and odontogenic keratocyst: correlation between contrast
enhancement and angiogenesis. J Comput Assist Tomogr 2002; 26: 922–6.
10. Jinzaki M, Tanimoto A, Mukai M et al. Double-phase helical CT of small renal
parenchymal neoplasms: correlation with pathologic findings and tumor
angiogenesis. J Comput Assist Tomogr 2000; 24: 835–42.
11. Ford JM, Miles KA, Hayball MP et al. A simplified technique for measurement
of blood-brain barrier permeability using computed tomography: preliminary
results of the effect of RMP-7. In: Faulkner K, Carey B, Crellin A, Harrison RM,
eds. Quantitative Imaging in Oncology. Proceedings of the 19th LH Gray
Conference. London: British Institute of Radiology 1996: 1–3.
12. Warren K, Jakacki R, Widemann B et al. Phase II trial of intravenous lobradimil
and carboplatin in childhood brain tumors: a report from the Children’s
Oncology Group. Cancer Chemother Pharmacol 2006; 58: 343–7.
13. Hockel M, Schlenger K, Mitze M et al. Hypoxia and radiation response in
human tumors. Semin Radiat Oncol 1996; 6: 3–9.
14. Nordsmark M, Overgaard M, Overgaard J. Pretreatment oxygenation predicts
radiation response in advanced squamous cell carcinoma of the head and
neck. Radiother Oncol 1996; 41: 31–9.
15. Falk SJ, Ramsay JR, Ward R et al. BW12C perturbs normal and tumor tissue
oxygenation and blood flow in man. Radiother Oncol 1994; 32: 210–17.
Media 1997: 89–116.
17. Mankoff DA, Dunnwald LK, Gralow JR et al. Changes in blood flow and
metabolism in locally advanced breast cancer treated with neoadjuvant
chemotherapy. J Nucl Med 2003; 44: 1806–14.
18. Tozaki M, Uno S, Kobayashi T et al. Histological breast cancer extent after
neoadjuvant chemotherapy: comparison with multidetector-row CT and
dynamic MRI. Radiat Med 2004; 22: 246–53.
9781842143094-Ch12 8/8/07 12:07 PM Page 213
19. Gandhi D, Chepeha DB, Miller T et al. Correlation between initial and early
follow-up CT perfusion parameters with endoscopic tumor response in
patients with advanced squamous cell carcinomas of the oropharynx treated
with organ-preservation therapy. Am J Neuroradiol 2006; 27: 101–6.
20. Choi J-B, Park C-K, Park DW et al. Does contrast enhancement on CT suggest
tumor response for chemotherapy in small cell carcinoma of the lung?
21. Kiessling F, Boese J, Corvinus C et al. Perfusion CT in patients with advanced
bronchial carcinomas: a novel chance for characterisation and treatment
monitoring? Eur Radiol 2004; 14: 1226–33.
22. Xiong HQ, Herbst R, Faria SC et al. A phase I surrogate endpoint study of
SU6668 in patients with solid tumors. Invest New Drugs 2004; 22: 459–66.
23. Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000;
407: 249–57.
24. Meijerink MR, van Cruijsen H, Hoekman K et al. The use of perfusion CT for
the evaluation of therapy combining AZD2171 with gefitinib in cancer patients.
Eur Radiol 2006 Oct 27; [EPub ahead of print].
specific antibody bevacizumab has antivascular effects in human rectal cancer:
Nat Med 2004; 10: 145–7.
26. McNeel DG, Eickhoff J, Lee FT et al. Phase I trial of a monoclonal antibody
specific for integrin (MEDI-522) in patients with advanced malignancies,
including an assessment on tumor perfusion. Clin Cancer Res 2005; 11: 7851–60.
27. Thomas JP, Arzoomanian RZ, Alberti D et al. Phase I pharmacokinetic and
pharmaocodynamic study of recombinant human endostatin in patients with
advanced solid tumors. J Clin Oncol 2003; 21: 223–31.
28. Winkler F, Kozin SV, Tong RT et al. Kinetics of vascular normalisation by
VEGFR2 blockade governs brain tumor response to radiation: role of oxygen-
tation, angiopoietin-1 and matrix metalloproteases. Cancer Cell 2004; 6: 553–63.
29. Harvey CJ, Blomley MJ, Dawson P, et al. Functional CT imaging of the acute
hyperaemic response to radiation therapy of the prostate gland: early experi-
ence. J Comput Assist Tomogr 2001; 25: 43–9.
30. Harvey C, Dooher A, Morgan J et al. Imaging of tumor therapy responses by
Radiology 2005; 234: 785–92.
32. Bondestam S, Halavaara JT, Jaaskelainen JE et al. Perfusion CT of the brain in
the assessment of flow alterations during brachytherapy of meningioma. Acta
Radiol 1999; 40: 469–73.
33. Tsushima Y, Funabasama S, Aoki J, Sanada S, Endo K. Quantitative perfusion
map of malignant liver tumors created from dynamic computed tomography
data. Acad Radiol 2004; 11: 215–23.
34. Purdie TG, Sherar MD, Lee TY. The use of CT perfusion to monitor the effect
of hypocapnia during laser thermal therapy in a rabbit model. Int J
Hypertherm 2003; 19: 461–79.
35. Sommerfeld NWB, Miles KA, Dugdale P, Leggett DAC, Bunce IH. Colorectal
cancer: progressive disease is associated with altered liver perfusion on functional
CT. Abstract presented at 50th Annual Meeting of the Royal Australian and
New Zealand College of Radiologists, 21–25 October, 1999, Sydney: Abstr No 56.
9781842143094-Ch12 8/8/07 12:07 PM Page 214
mined noninvasively by dynamic computed tomography predicts outcome in
head-and-neck cancer after radiotherapy. Int J Radiat Oncol Biol Phys 2003;
57: 1351–6.
37. Millar BM, Purdie TG, Yeung I et al. Assessing perfusion changes during
whole brain irradiation for patients with cerebral metastases. J Neurooncol
2005; 71: 281–6.
38. Miles KA; Hayball MP, Dixon AK. Functional imaging of changes in human
intra-renal perfusion using quantitative dynamic computed tomography. Invest
Radiol 1994; 29: 911–14.
39. Mulder WJ, Strijkers GJ, Habets JW et al. MR molecular imaging and fluores-
cence microscopy for identification of activated tumor endothelium using a
bimodal lipidic nanoparticle. FASEB J 2005; 19: 2008–10.
40. Wang ZQ, Li JS, Lu GM et al. Correlation of CT enhancement, tumor angio-
genesis and pathologic grading of pancreatic carcinoma. World J Gastroenterol
2003; 9: 2100–4.
9781842143094-Ch13 8/8/07 12:08 PM Page 215
13
Perfusion CT–PET:
opportunities for combined
assessment of tumor
vascularity and metabolism
Kenneth A Miles
Technology that integrates computed tomography (CT) and positron

emission tomography (PET) within a single imaging device is a rela-
tively recent development in imaging. Nevertheless, such systems are
fast becoming the standard for delivery of clinical PET services. The CT
component of these combined examinations has generally comprised
an examination for attenuation correction and anatomical assignment of
abnormalities identified on PET, and is therefore performed without
contrast medium. However, there is a growing interest in the use of
intravenous contrast media during PET–CT.1 Extending these applica-
tions for contrast media to include CT perfusion would enable anatom-
ical information about tumors to be co-registered with perfusion data
and metabolic information, such as glucose metabolism, in a single
examination. This combined approach would allow simultaneous
assessment of multiple endothelial-related (i.e. blood flow, blood
volume, vascular permeability) and tumor cell-related (i.e. metabolism)
aspects of tumor biology (Figure 13.1). The use of CT to assess perfu-
sion as opposed to administration of a second PET tracer such as [O15]-
water circumvents the need for an on-site cyclotron. Furthermore,
because CT depicts perfusion data with higher spatial resolution, some
of the limitations of PET perfusion studies can be avoided, including the
underestimation of perfusion values in small tumors due to the partial
volume effect, and the spill over of counts from adjacent structures with
high blood flow (e.g. heart, aorta, liver).2 To date, there have been few
reports describing the implementation of CT perfusion and PET on a
9781842143094-Ch13 8/8/07 12:08 PM Page 216
a b
c d
Figure 13.1 Combining fluorodeoxyglucose-positron emission tomography

(FDG-PET) with computed tomography perfusion (CT) allows simultaneous
assessment of anatomy (a: conventional CT), glucose metabolism (b: FDG-PET),
and multiple endothelial related parameters (c: relative blood volume, d: vascu-
lar permeability). The recurrent glioma is seen as focally increased glucose
metabolism and increased vascular permeability (arrows)
single imaging device. However, there have been a number of studies

in which patients have been examined by both techniques on separate
systems. These studies, combined with data in which other methods of
perfusion imaging have been correlated against PET, have shown that
tumor vascularity and metabolism are not consistently coupled, and a
combined assessment using CT perfusion and PET in a single examina-
tion could provide a more detailed picture of the biological behavior of
tumors.
9781842143094-Ch13 8/8/07 12:08 PM Page 217
Perfusion CT–PET 217
BIOMOLECULAR MEDIATORS FOR TUMOR VASCULARITY

AND METABOLISM
The tumor circulation is an important determinant of the microenviron-

ment within tumor cells, including their metabolic status. A well devel-
oped tumor vascular supply will ensure a sufficient delivery of glucose
and oxygen to support the metabolism essential for tumor growth.
However, as tumors increase in size, they may outgrow their blood
supply. The subsequent reduction in the delivery of oxygen renders
such tumors hypoxic. Yet, hypoxia is also recognized to be a feature of
the tumor microenvironment that stimulates glucose metabolism.3,4
Thus, the relationship between tumor vascularity and metabolism is
complex.
Tumor hypoxia is known to be associated with increased tumor
aggression and a poor response to a variety of therapeutic strategies.3
Hypoxia inducible factors (HIFs), such as HIF-1α and HIF-2α, are
important mediators of the tumor metabolic response to hypoxia. HIFs
increase the transcription of several molecules that adapt the tumor to
its hypoxic environment. Not only do HIFs increase the expression of
glucose transporters and hexokinase, HIFs also stimulate angiogenesis,
promote tumor aggression, and confer resistance to chemotherapy and
radiotherapy (Figure 13.2).3 The ability for tumors to mount an effec-
tive response to hypoxia is highly variable. Histological and imaging
Oncogene
Tissue mutations
hypoxia e.g. p53
HIF
Upregulated
Nitric oxide VEGF Glut-1 MDR p-Gp
molecules
Vascular
Biological Increased Multidrug
dilatation and Angiogenesis
effects glucose metabolism resistance
permeability
Figure 13.2 Summary of the effects of hypoxia inducible factors (HIFs) on

tumor biology. VEGF, vascular endothelial growth factor; Glut-1, glucose trans-
porter; MDR P-Gp, multidrug resistance P-glycoprotein
9781842143094-Ch13 8/8/07 12:08 PM Page 218
studies that compare markers of hypoxia and glucose metabolism have

shown that these aspects of the tumor microenvironment do not always
correlate.5–7
HIFs can also be expressed constitutively by tumors as a conse-
quence of oncogene mutations, resulting not only in increased glucose
metabolism but also increased angiogenesis and perfusion.3 Indeed,
high levels of angiogenesis and elevated glucose metabolism are both
associated with increased metastatic potential and poor patient survival
for a range of cancers.8–15 However, in non-small-cell lung cancer,
increased expression of HIFs with low microvessel density (MVD) has
also been shown to be associated with a worse prognosis.16 Thus, the
balance between tumor vascularity and metabolic status offers impor-
tant information concerning the tumor microenvironment. High glucose
metabolism with increased vascularity would represent a different bio-
logical status of the tumor than high metabolism with poor vascularity,
the latter indicating adaptation to hypoxia. Low glucose metabolism
with poor vascularity would suggest a failure of the adaptive response
to hypoxia.
VASCULAR–METABOLIC RELATIONSHIPS IN TUMORS
Imaging studies using a range of techniques suggest a highly variable rela-

tionship between tumor vascularity and glucose metabolism (Table 13.1).
Factors influencing the relationship include tumor type, grade and size.
Although demonstrating an overall trend between glucose metabolism
and blood volume in cerebral glioma, Aronen et al found that uncou-
pling of these processes could occur with high-grade tumors.17 Studies
in breast cancer have found only poor or moderate correlations
between vascularity and metabolism,18,19 whilst studies of liver tumors
have suggested a negative correlation to date20,21 (Figure 13.3). In non-
small-cell lung cancer (NSCLC), the relationship between tumor circu-
lation and metabolism appears to be dependent on tumor size. One
study comprising patients with NSCLC undergoing surgical resection
found a statistically significant correlation between fluorodeoxyglucose
(FDG) uptake and CT measurements of tumor vascularity.22 The mean
diameter of the tumors included in this study was 2.9 ± 0.3 cm. On the
other hand, a separate study comparing PET measurements of perfu-
sion with FDG uptake in patients with stage IIIA-N2 NSCLC, and thus
probably larger tumors, showed no correlation.23 A more recent study
using CT perfusion and FDG-PET to specifically assess the impact of
tumor size on the relationship between blood flow and metabolism
9781842143094-Ch13 8/8/07 12:08 PM Page 219
Table 13.1 Summary of imaging studies comparing tumor vascularity and metabolism
Study Tumor type Techniques Findings
Aronen et al.17 Glioma DC-MRI Maximum CBV correlates with

FDG-PET maximum FDG (r = 0.573, p = 0.023)
Mankoff et al.18 Breast H215O-PET Perfusion and metabolism weakly
FDG-PET correlated. High metabolism–flow
ratio predicts poor treatment
response
Semple et al.19 Breast DC-MRI Moderate correlation between
FDG-PET vascularity and metabolism
Fukuda et al.20 HCC, CCC, H215O-PET Negative correlation (r = −0.713,
colon liver FDG-PET p = 0.006)
mets
Van Laarhoven Colon liver DC-MRI Negative trend (r = −0.421,
et al.21 mets FDG-PET p = 0.082)
Tateishi et al.22 NSCLC Perf CT Vascularity and metabolism correlate
FDG-PET in surgically resectable tumors
Hoekstra et al.23 NSCLC H215O-PET No correlation between perfusion and
FDG-PET metabolism in stage IIIA-N2
Miles et al.24 NSCLC Perf CT Correlation between vascularity and
FDG-PET metabolism in small tumors
only (r = 0.85, p = 0.03)
Veronesi et al.25 Lung Perf CT FDG uptake and angiogenesis
metastases FDG-PET independent
CBV, cerebral blood volume; CCC, cholangiocarcinoma; DC-MRI, dynamic contrast-enhanced magnetic
resonance imaging; FDG-PET, fluorodeoxyglucose-positron emission tomography; HCC, hepatocellular
carcinoma; H215O, oxygen-15-labeled water; mets, metastases; NSLLC, non-small-cell lung cancer;
Perf CT, computed tomography perfusion.
found a correlation between these parameters only for tumors with a

cross-sectional area of less than 4.5 cm2, equivalent to a mean diame-
ter of 2.4 cm.24 Larger tumors tended to exhibit lower perfusion values
but greater FDG uptake. The metabolic–flow difference correlated with
tumor size, implying uncoupling of blood flow and glucose metabolism
in larger tumors. Uncoupling of flow and metabolism has also been
observed in pulmonary metastases.25 The degree to which such uncou-
pling of flow and metabolism develops in larger tumors of other cancer
types has yet to be determined.
The dependence of blood flow–metabolic relationships in NSCLC on
tumor size has implications for the use of FDG-PET and quantitative
CT perfusion in the evaluation of pulmonary nodules.26 The finding
of low perfusion in large tumors highlights an important pitfall for CT
perfusion in the diagnostic assessment of pulmonary nodules, particu-
larly for nodules with a diameter greater than approximately 2.5 cm.
9781842143094-Ch13 8/8/07 12:08 PM Page 220
Figure 13.3 Corresponding conventional contrast-enhanced CT (a), CT perfusion

(b), and FDG-PET (c) images demonstrating increased perfusion and increased
glucose metabolism in hepatic metastases from colorectal cancer
9781842143094-Ch13 8/8/07 12:08 PM Page 221
Histological studies in lung carcinomas indicate this pitfall to be due to

an association between tumor necrosis and lower contrast enhance-
ment [27]. On the other hand, vascularity well in excess of metabolism
may be a feature of inflammatory lesions (Figure 13.4)28
Vascular–metabolic relationships and tumor aggression

As discussed above, tumor adaptation to hypoxia is associated with
increased tumor aggression and resistance to treatment. The finding of
high FDG uptake with poor vascularity offers a potential imaging
correlate for this tumor biological status. Indeed, Miles et al.24 found
that a high metabolic–flow difference was more likely to occur in
advanced NSCLC, whilst Mankoff et al.18 showed that breast cancers
with a high ratio of glucose metabolism to perfusion were less likely to
respond favorably to treatment. In separate imaging studies of head and
neck cancer, low perfusion assessed by CT perfusion and high FDG
uptake have both been found to be independent predictors for poor
local control following treatment.29,30 However, the significance of
SPV = 5.5 10 SUV = 6.3
0
SPV = 5.5 20 SUV = 2.8
a b 0 c
Figure 13.4 Conventional CT (a), CT perfusion (b), and FDG-PET (c) images
of malignant (upper row) and inflammatory (lower row) lung lesions. In the
malignant lesion, glucose metabolism (expressed as the standardized uptake
value (SUV)), exceeds perfusion (expressed as the standardized perfusion value
(SPV)), whereas the flow–metabolic relationship is reversed in the inflammatory
lesion
9781842143094-Ch13 8/8/07 12:08 PM Page 222
vascular–metabolic relationships may be tumor-specific. In separate

studies of rectal cancer, high tumor perfusion before chemoradiation,
assessed with dynamic contrast-enhanced MRI or CT perfusion, pre-
dicted a poor response,31,32 whilst pretreatment tumor uptake of FDG
showed no correlation with tumor shrinkage rate, presence of micro-
scopic residual disease, or local recurrence rate.33
The emergence of chemotherapeutic agents, such as tirapazamine,
and gene therapy strategies that target hypoxia suggest a clinical role
for the imaging detection of tumor hypoxia.3 Certain radiopharmaceu-
ticals such as [18F]fluoromisonidazole (F-MISO) are selectively taken up
in proportion to tissue hypoxia. However, such imaging agents provide
little information about the ability of tumor cells to adapt to hypoxia,
for example by increasing glycolysis. Imaging assessments of the
balance between glucose metabolism and perfusion will more closely
reflect tumor adaptation to hypoxia, and are thus potentially of greater
prognostic significance than the presence of hypoxia itself.
Changes in tumor vascularity and metabolism

following therapy
Imaging is widely used to assess tumor response to therapy, both in
clinical practice and in trials assessing the efficacy of novel drugs or
other therapies. The main imaging approaches used are predicated on
serial measurements of tumor size on CT or other cross-sectional
imaging techniques (e.g. RECIST: Response Evaluation Criteria in Solid
Tumors). However, such anatomically based methods are constrained
by poor reproducibility, slow response rates, and residual non-tumor-
ous masses. Furthermore, certain treatment approaches, for example
drugs that target the tumor vasculature, may produce little or no change
in tumor size despite therapeutic efficacy. In the light of these limita-
tions, there is increasing interest in using imaging techniques that assess
changes in tumor physiology in response to therapy, rather than change
in size. As discussed in Chapter 12, CT perfusion is one such approach,
but the use of FDG-PET as a tumor response marker is also emerging.
To date, these methods have mostly been used independently.
However, the limited data available in which both perfusion and FDG
uptake have been measured before and after treatment show that per-
fusion and glucose metabolism may not change in parallel in response
to therapy, reflecting different responses of the vascular and cellular
compartments of the tumor34–37 (Figure 13.5). Tumor type, drug type
and dose, and time since therapy are all factors that affect the relative
magnitude of change in these parameters.
9781842143094-Ch13 8/8/07 12:08 PM Page 223
Before
therapy
Tumor area: Perfusion:

Glucose metabolism:
−85% and −83% −12% and +3%
−42% and −68%
After
therapy
CT Perfusion CT FDG-PET
Figure 13.5 Different morphological, vascular, and metabolic responses to

therapy in non-small-cell lung cancer. (Reproduced with permission from
reference 38)
The studies that have assessed changes in both tumor perfusion

and glucose metabolism following therapy to date are summarized in
Table 13.2. In many cases, the reduction in perfusion is of greater mag-
nitude than the fall in glucose metabolism. Uncoupling of flow and
metabolism appears to be particularly likely following antivascular ther-
apy, probably reflecting drug-induced hypoxia and secondary stimula-
tion of glucose metabolism. Thus, the relative magnitude of changes in
tumor perfusion and glucose metabolism in response to therapy may
depend upon both tumor type and treatment regimen. The study of
rectal cancer by Willett et al. demonstrating different CT perfusion and
FDG-PET responses to the vascular endothelial growth factor (VEGF)
antibody bevacizumab particularly highlights the benefits of combining
CT perfusion with FDG-PET.37
Based on the results of above studies, it is possible to propose a sub-
classification of therapeutic responses into those that are: (1) balanced
(i.e. a significant reduction in both glucose metabolism and tumor vas-
cularity), (2) predominantly vascular, and (3) predominantly metabolic
(Table 13.3). A balanced response seems most likely to be associated
with a good outcome. It is likely that the predominantly vascular and
predominantly metabolic responses will carry different clinical signifi-
cance. The possibility of modulating tumor responses by adapting ther-
apy for individual patients on the basis of their imaging findings can be
envisaged. In the study of Willett et al. the addition of chemoradiation
to bevacizumab produced a more balance response, and this anti-VEGF
agent is currently licensed for use in combination with other therapies.
224
9781842143094-Ch13
8/8/07
Table 13.2 Summary of studies assessing changes in both tumor perfusion and glucose metabolism following therapy
Study Tumor type Drug regimen Techniques Findings

12:08 PM
Herbst et al.34 Various Endostatin H215O-PET Flow–metabolic relationships dose–dependent. Reduced

FDG-PET perfusion but increased glucose metabolism at dose that
produced greatest endothelial and tumor cell apoptosis
Mankoff et al.35 Locally advanced Neoadjuvant H215O-PET Increased perfusion with small reduction in FDG uptake
Page 224
breast cancer chemotherapy FDG-PET in tumors failing to respond to treatment. Decreased

perfusion and FDG uptake in tumors that proceeded to
partial or complete response
Multidetector computed tomography in oncology
Kurdziel et al.36 Androgen-independent Thalidomide H215O-PET Positive correlation between prostate specific antigen (PSA)
prostate cancer FDG-PET response and change in glucose metabolism. Negative
correlation between PSA response and change
in perfusion
Willett et al.37 Rectal Bevacizumab Perf CT Significant reduction in perfusion. No change in FDG
cancer FDG-PET uptake
9781842143094-Ch13 8/8/07 12:08 PM Page 225
Table 13.3 Sub–classification of functional tumor response based upon perfusion

and metabolic imaging
Unchanged or increased Reduced metabolism

metabolism
Unchanged or increased No response Predominantly metabolic

perfusion response
Reduced perfusion Predominantly vascular Balanced response
response
On the other hand, it may be appropriate to add an antivascular drug

to a treatment regimen producing a predominantly metabolic response.
The ultimate goal would be to tailor an individual patient’s therapy to
the vascular–metabolic response exhibited by their tumor.
HEPATIC PHOSPHORYLATION OF GLUCOSE ASSESSED BY

CT PERFUSION–PET
FDG-PET has been used to evaluate the kinetics of liver glucose metab-
olism using a three-compartment model with four rate constants (k1–4)
that define the rates of transfer between plasma FDG, tissue FDG, and
tissue FDG-6-phosphate (Figure 13.6).39 CT measurements of hepatic
perfusion can feed into this kinetic analysis by providing an estimate
of the rate of passage between intravascular and intracellular non-
phosphorylated compartments, k1.40 The permeability of the liver sinu-
soids to glucose is sufficiently high for this rate constant approximate
perfusion, as confirmed by the studies of Munk et al.41 Hence, the ratio
of hepatic FDG uptake to hepatic perfusion provides an index of the
hepatic phosphorylation fraction (HPFI: see Box 13.1).
Applying the above methodology to patients with colorectal cancer
it has been possible to demonstrate reduced hepatic phosphorylation of
k1 k3
Plasma Tissue Tissue
FDG FDG FDG-6-PO4−
k2 k4
Figure 13.6 Three-compartment kinetic model for hepatic glucose metabolism.39

FDG, fluorodeoxyglucose; FDG-6-PO4-, FDG-6-phosphate
9781842143094-Ch13 8/8/07 12:08 PM Page 226
Box 13.1 Derivation of the hepatic phosphorylation fraction index (HPFI)
Using the three-compartmental model described by Choi et al. (Figure 13.6),39 the fraction
of intracellular glucose that undergoes phosphorylation (i.e. the phosphorylation fraction,
PF) is given by:
PF = k3/(k2 + k3) (1)
The net influx constant, Ki is given by:
Ki = (k1k3)/(k2 + k3) = k1 × PF (2)
If the standardised uptake value (SUV) of FDG in the liver is used as a surrogate for the
net influx constant (Ki) and CT perfusion measurements of combined arterial and portal
perfusion as a surrogate for k1, then an index of hepatic phosphorylation (HPFI) can be
obtained from:
HPFI = SUV / Hepatic perfusion (3)
glucose in patients with extrahepatic tumors (HPFI 2.62 vs. 4.05,

p < 0.002) and in those patients surviving less than 2 years (HPFI 2.84
vs. 4.06, p = 0.014).40 There was no evidence of weight loss amongst
the patients with reduced phosphorylation.
Derangements of liver metabolism in patients with cancer are well
recognized, including reduced hepatic glucose production and glucose
recycling and altered glucose transport across hepatic cell membranes.42
These abnormalities contribute to the syndrome of anorexia and weight
loss known as cachexia, a common outcome for patients with advanced
cancer. Cachexia is a cause of significant morbidity, affecting both
the efficacy of antitumor therapy and patient survival, and has there-
fore been identified as a target for therapy.43,44 CT perfusion–PET
assessments of hepatic glucose phosphorylation offer the potential to
identify metabolic changes associated with cachexia before the devel-
opment of weight loss and to provide a means of monitoring treatments
aimed at controlling those metabolic abnormalities.
SUMMARY
The increased glucose metabolism exhibited by tumors has been

recognized since Warburg’s experiments in the 1930s.45 The fact that
the growth, survival, and expansion of solid tumors are highly depend-
ent on the vascular system recruited by the malignancy was demon-
strated in the early 1970s by Folkman.46 Although it is reasonable
to hypothesize that the metabolic requirements of tumors are mirrored
9781842143094-Ch13 8/8/07 12:08 PM Page 227
by alterations in tumor hemodynamics, experience has shown a much

more complex relationship between these processes. The balance
between vascularity and metabolism is an important determinant of
the biological behavior of tumors and their response to therapy. CT
perfusion–PET using integrated imaging devices now offers the oppor-
tunity to quantify both these processes in-vivo in a single examination,
not only for research but also in the clinical arena.
REFERENCES
1. Antoch G, Freudenberg LS, Beyer T, Bockisch A, Debatin JF. To enhance or

not to enhance? 18F-FDG and CT contrast agents in dual-modality 18F-FDG
PET/CT. J Nucl Med 2004; 45(Suppl) 1: 56S–65S.
2. Bacharach SL, Libutti SK, Carrasquillo JA. Measuring tumor blood flow with
H(2)(15)O: practical considerations. Nucl Med Biol 2000; 27: 671–767.
3. Semenza GL. HIF-1 and tumor progression: pathophysiology and therapeutics.
Trends Mol Med 2002; 8(4 Suppl): S62–7.
4. Raghunand N, Gatenby RA, Gillies RJ. Microenvironmental and cellular conse-
quences of altered blood flow in tumors. Br J Radiol 2003; 76: 11–22.
5. van Laarhoven HW, Kaanders JH, Lok J et al. Hypoxia in relation to vascula-
ture and proliferation in liver metastases in patients with colorectal cancer. Int
J Radiat Oncol Biol Phys 2006; 64: 473–82.
6. Foo SS, Abbott DF, Lawrentschuk N, Scott AM. Functional imaging of intra-
tumoral hypoxia. Mol Imaging Biol 2004; 6: 291–305.
7. Rajendran JG, Wilson DC, Conrad EU et al. [(18)F]FMISO and [(18)F]FDG PET
imaging in soft tissue sarcomas: correlation of hypoxia, metabolism and VEGF
expression. Eur J Nucl Med Mol Imaging 2003; 30: 695–704.
8. Ahuja V, Coleman RE, Herndon J, Patz EF Jr. The prognostic significance of
fluorodeoxyglucose positron emission tomography imaging for patients with
non-small cell lung carcinoma. Cancer 1998; 83: 918–24.
9. Vansteenkiste JF, Stroobants SG, Dupont PJ et al. Prognostic importance of the
standardized uptake value on (18)F-fluoro-2-deoxy-glucose-positron emission
tomography scan in non-small-cell lung cancer: an analysis of 125 cases.
Leuven Lung Cancer Group. J Clin Oncol 1999; 17: 3201–6.
10. Higashi K, Ueda Y, Arisaka Y et al. 18F-FDG uptake as a biologic prognostic
factor for recurrence in patients with surgically resected non-small cell lung
cancer. J Nucl Med 2002; 43: 39–45.
11. Jeong HJ, Min JJ, Park JM et al. Determination of the prognostic value of [(18)F]
fluorodeoxyglucose uptake by using positron emission tomography in patients
with non-small cell lung cancer. Nucl Med Commun 2002; 23: 865–70.
12. Downey RJ, Akhurst T, Gonen M et al. Preoperative F-18 fluorodeoxyglucose-
positron emission tomography maximal standardized uptake value predicts
survival after lung cancer resection. J Clin Oncol 2004; 22: 3255–60.
13. Fontanini G, Bigini D, Vignati S et al. Microvessel count predicts metastatic
disease and survival in non-small cell lung cancer J Pathol 1995; 177:
57–63.
9781842143094-Ch13 8/8/07 12:08 PM Page 228
14. Meert AP, Paesmans M, Martin B et al. The role of microvessel density on the
survival of patients with lung cancer: a systematic review of the literature with
meta-analysis. Br J Cancer 2002; 87: 694–701.
15. Tanaka F, Yanagihara K, Otake Y, et al. Prognostic factors in resected patho-
logic (p-) stage IIIA-N2, non-small-cell lung cancer. Ann Surg Oncol 2004; 11:
612–18.
16. Giatromanolaki A, Koukourakis MI, Sivridis E et al. Relation of hypoxia
inducible factor 1 alpha and 2 alpha in operable non-small cell lung cancer
to angiogenic/molecular profile of tumors and survival. Br J Cancer 2001;
85: 881–90.
17. Aronen HJ, Pardo FS, Kennedy DN et al. High microvascular blood volume is
associated with high glucose uptake and tumor angiogenesis in human
gliomas. Clin Cancer Res 2000; 6: 2189–200.
18. Mankoff DA, Dunnwald LK, Gralow JR et al. Blood flow and metabolism in
locally advanced breast cancer: relationship to response to therapy. J Nucl Med
2002; 43: 500–9.
19. Semple SI, Gilbert FJ, Redpath TW et al. The relationship between vascular
and metabolic characteristics of primary breast tumors. Eur Radiol 2004; 14:
2038–45.
20. Fukuda K, Taniguchi H, Koh T, Kunishima S, Yamagishi H. Relationships
between oxygen and glucose metabolism in human liver tumors: positron
emission tomography using (15)O and (18)F-deoxyglucose. Nucl Med
Commun 2004; 25: 577–83.
21. van Laarhoven HW, de Geus-Oei LF, Wiering B et al. Gadopentetate dimeglu-
mine and FDG uptake in liver metastases of colorectal carcinoma as deter-
mined with MR imaging and PET. Radiology 2005; 237: 181–8.
22. Tateishi U, Nishihara H, Tsukamoto E et al. Lung tumors evaluated with FDG-
PET and dynamic CT: the relationship between vascular density and glucose
metabolism. J Comput Assist Tomogr 2002; 26: 185–90.
23. Hoekstra CJ, Stroobants SG, Hoekstra OS et al. Measurement of perfusion in
stage IIIA-N2 non-small cell lung cancer using H(2)(15)O and positron emis-
sion tomography. Clin Cancer Res 2002; 8: 2109–15.
24. Miles KA, Griffiths MR, Keith CJ. Blood flow-metabolic relationships are
dependent on tumor size in non-small cell lung cancer: a study using quanti-
tative contrast-enhanced computer tomography and positron emission tomog-
raphy. Eur J Nucl Med Mol Imaging 2006; 33: 22–8.
25. Veronesi G, Landoni C, Pelosi G et al. Fluoro-deoxi-glucose uptake and angio-
genesis are independent biological features in lung metastases. Br J Cancer
2002; 86: 1391–5.
27. Yamashita K, Matsunobe S, Tsuda T et al. Intratumoral necrosis of lung carci-
noma: a potential diagnostic pitfall in incremental dynamic computed tomogra-
phy analysis of solitary pulmonary nodules? J Thorac Imaging 1997; 12: 181–7.
Radiology 2001; 220: 548–53.
mined noninvasively by dynamic computed tomography predicts outcome in
9781842143094-Ch13 8/8/07 12:08 PM Page 229
head-and-neck cancer after radiotherapy. Int J Radiat Oncol Biol Phys 2003;
57: 1351–6.
30. Allal AS, Slosman DO, Kebdani T et al. Prediction of outcome in head-and-
neck cancer patients using the standardized uptake value of 2-[18F]fluoro-2-
deoxy-D-glucose. Int J Radiat Oncol Biol Phys 2004; 5: 1295–300.
31. Devries AF, Griebel J, Kremser C et al. Tumor microcirculation evaluated by
dynamic magnetic resonance imaging predicts therapy outcome for primary
rectal carcinoma. Cancer Res 2001; 61: 2513–16.
Radiology 2005; 234: 785–92.
33. Oku S, Nakagawa K, Momose T et al. FDG-PET after radiotherapy is a good
prognostic indicator of rectal cancer. Ann Nucl Med 2002; 16: 409–16.
34. Herbst RS, Mullani NA, Davis DW et al. Development of biologic markers of
response and assessment of antiangiogenic activity in a clinical trial of human
recombinant endostatin. J Clin Oncol 2002; 20: 3804–14.
35. Mankoff DA, Dunnwald LK, Gralow JR et al. Changes in blood flow and
metabolism in locally advanced breast cancer treated with neoadjuvant
chemotherapy. J Nucl Med 2003; 44: 1806–14.
36. Kurdziel KA, Figg WD, Carrasquillo JA et al. Using positron emission tomog-
raphy 2-deoxy-2-[18F]fluoro-D-glucose, 11CO, and 15O-water for monitoring
androgen independent prostate cancer. Mol Imaging Biol 2003; 5: 86–93.
Nat Med 2004; 10: 145–7.
38. Miles KA. Computed tomography measurements of perfusion in cancer therapy.
In: Shields AF, Price P, eds. In vivo Imaging of Cancer Therapy Totowa, NJ:
Humana Press, 2007: in press.
39. Choi Y, Hawkins RA, Huang SC et al. Evaluation of the effect of glucose inges-
tion and kinetic model configurations of FDG in the normal liver. J Nucl Med
1994; 35: 818–23.
40. Miles KA, Comber L, Keith CJ et al. Combining perfusion CT and FDG-PET
data demonstrates reduced hepatic phosphorylation of glucose in patients with
advanced colorectal cancer and poor survival. Eur Radiol 2003; 13(Suppl 1): 293.
41. Munk OL, Bass L, Roelsgaard K et al. Liver kinetics of glucose analogs
measured in pigs by PET: importance of dual-input blood sampling. J Nucl
Med 2001; 42: 795–801.
42. Tayek JA. A review of cancer cachexia and abnormal glucose metabolism in
humans with cancer. J Am Coll Nutr 1992; 11: 445–56.
43. Capra S, Ferguson M, Ried K. Cancer: impact of nutrition intervention out-
come—nutrition issues for patients. Nutrition 2001; 17: 769–72.
44. Laviano A, Meguid MM, Inui A, Muscaritoli M, Rossi-Fanelli F. Therapy insight:
cancer anorexia-cachexia syndrome—when all you can eat is yourself. Nat
Clin Pract Oncol 2005; 2: 158–65.
45. Warburg O, Wind F, Negelin E. The metabolism of tumors in the body. J Gen
Physiol 1927; 8: 519–30.
285: 1182–6.
9781842143094-Ch13 8/8/07 12:08 PM Page 230
9781842143094-Idx 8/8/07 5:37 PM Page 231
Index
Page numbers in italics indicate figures, tables or boxes.
accuracy studies 43 appearance (arrival) time (T0)

adiabatic approximation 30–1 34, 62
Ambrose, J. 7, 8 maps 61–2, 63
anatomy Aristotle 3, 4
history of development 3–6 arrival time (T0) see appearance
imaging and 6–8 time
knowledge in ancient times 1–3 arterial concentration function,
ancient civilizations 1–3 input 31–6
angiogenesis 73–83, 202 delay in contrast agent arrival
characteristics 76–7 33–4
endogenous inhibitors 75 dispersion 33
imaging 77–80, 82–3 to liver 34–5
rectal cancer 129, 134–6 recirculation effect 35–6
therapeutic inhibitors see attenuation
antiangiogenic agents change over time 55, 56, 56
in tumor growth, progression thresholds for segmentation 66,
and metastases 74–5 66, 67
angiogenic switch 74–5 tube voltage and 54, 54
angiopoietins 82 Avastin see bevacizumab
animal tumor models, validation Axel, Leon 11
studies 39–41, 42 axillary lymph nodes, sentinel
annexin V 82 165, 165
anthracycline 200, 203 AZD21271 202–4, 203
antiangiogenic agents 73–4, 82, 198
cancer management 77–8 beam hardening effects 38, 54
evaluation of efficacy 78–81, 202–5 lungs 116, 123
lymphoma 169 lymph nodes 164
rectal cancer 137–9 benign prostatic hyperplasia
antivascular agents, novel 201–5 148–50, 149
9781842143094-Idx 8/8/07 5:37 PM Page 232
232 Index
bevacizumab 73–4, 77, 81 breast cancer

perfusion CT Perfusion 138–9, combined perfusion CT
202, 203 Perfusion–PET 224
perfusion CT–PET 223, 224 lymph node metastases 164,
binomial spatial smoothing 62–4, 165, 165
63, 64 treatment response assessment
blood–brain barrier (BBB) 203
permeability 11 vascularity–metabolism
brain tumors 92, 93, 95, 96 relationship 218, 219, 221
drugs affecting 198–9, 199 bronchogenic carcinoma 112
radiation planning 98, 98 BW12C 199–200, 200, 203
blood flow (perfusion) (F) 16, 62
average, in region of interest 64, cachexia 226
68, 70 calibration, CT system 38, 114,
central volume principle 19–20 115, 116
comparison with enhancement capillary endothelium, transfer flux
58, 58–9 of solute through 17–19, 18
maps 61, 63 capillary permeability surface area
reproducibility studies 42–3 product (PS) 16–17, 62
validation of measurements 39–41 average, in region of interest
blood volume (Vb; BV) 16, 62 68, 71
average, in region of interest maps 61, 63
64–5, 68, 70 reproducibility studies 42–3
central volume principle 19–20 validation of measurement
maps 61, 63 39–41, 42
reproducibility studies 42–3 carbon-11-labeled carbon
validation of measurement 40–1 monoxide 80
bone marrow transplantation, carcinoembryonic antigen 191
complications 209, 211 central volume principle
bowel preparation, rectal cancer 130 19–20, 181
brain tumors 89–100 cerebral angiography, history of
choice of protocol 57 development 6–7
combined perfusion CT cerebral blood flow (CBF)
Perfusion–PET 216 brain tumors 93, 97
development of perfusion deconvolution method 11
imaging 10, 11–12 Kety–Schmidt method 10, 26
history of imaging 7, 8, 10 cerebral blood volume (CBV) 93,
image segmentation 66, 66, 68 94, 95, 97
parameter maps 61, 63 radiation necrosis vs recurrent
radiation planning 97–8, 98 tumor 99
treatment response assessment cerebral infarction 96–7
99, 198–9, 199, 203 cerebral lymphoma 94–6
validation of measurements 40 cerebral tumors see brain tumors
vascularity–metabolism cervical cancer 158–9
relationship 218, 219 chemoradiotherapy
vs cerebral infarction 96–7, 97 predicting response to 207–8
9781842143094-Idx 8/8/07 5:37 PM Page 233
Index 233
chemoradiotherapy (Continued) contrast media, X-ray 15–16

response assessment 136–7, delayed arrival in tissue 33–4
137, 206 history of development 6–7
chemotherapy injection rates 55–6, 56
assessment of complications input arterial concentration
209, 210 function 31–6
predicting response to 207 kinetic models 26–31
response assessment 99, 168, plasma normalized
200, 200–1, 224 quantities 20
China, ancient 2 rectal cancer 131
circulation, historical discovery 8–10 signal-to-noise ratio 38
colorectal cancer type 55
antiangiogenic therapy 77 volume and concentration
combined perfusion CT 55–7, 56
Perfusion–PET 225–6 convolution 20, 22–4, 23, 178
factors predicting survival cooption, vascular 74
186, 187 Cormack, Allan McLeod 8
liver metastases 173, 184, 185, Crookes, Sir William 6
188–9, 192–3 CT see computed tomography
occult hepatic micrometastases CT enhancement see enhancement,
185–6 CT
prediction of response to CT number 20
therapy 207 CTP see perfusion computed
reproducibility studies 42 tomography
see also rectal cancer
compartment models 27, 27–9 deconvolution analysis 11, 31
contrast injection rate and liver metastases 185
55–6, 56 liver perfusion 178–83
hepatic glucose metabolism 225, protocols 49, 51, 53, 55
225, 226 theory 24, 178–80, 179
liver perfusion 177–8 dexamethasone 199, 203
complications of treatment, dispersion, measured arterial
assessing 209 concentration function 33
computed tomography (CT) dissection, human body 2, 3, 4–5
angiography 47 distributed parameter models
history of development 7–8 29–31
perfusion see perfusion distribution volume (Vd) 16
computed tomography fractional 181
system calibration 38, 114, liver 177, 182, 183
115, 116 Doppler perfusion index, liver 188–9
computer-aided diagnosis (CAD), Doppler ultrasound
lung cancer 124 liver metastases 188–9
contrast appearance time see prostate cancer 149
appearance (arrival) time dynamic contrast-enhanced
contrast enhancement 20, 55–7 magnetic resonance imaging
see also tumor enhancement (DCE-MRI) 79, 81, 189
9781842143094-Idx 8/8/07 5:37 PM Page 234
234 Index
Egypt, ancient 1–2 functional computed tomography

endostatin (Endostar) 74, 203, (CT) 79
204, 224
endothelial progenitor Galen, Claudius 2, 3–4, 5, 9
cells (EPCs) 75 gastric cancer, lymph node
enhancement, CT 20, 55–7 metastases 164, 164–5
see also tumor enhancement gastrointestinal stromal tumors
exposure time 53–5 (GIST) 78, 81
extraction efficiency gastrointestinal tumors 129–41
(fraction) (E) 17 treatment response
assessment 203
FE 19, 62 see also colorectal cancer; rectal
two-compartment model 28–9 cancer
validation of measurement gefitinib 202–4, 203
39–40, 42 Geynes, John 4
fibroblast growth factors (FGFs) glioblastoma 94
75, 202 gliomas 93–4, 94, 95
Fick, Adolf 9–10 combined perfusion CT-
Fick principle Perfusion–PET 216
application 12, 25, 27 vascularity–metabolism
history 9–10 relationship 218, 219
validation 39, 43 glucose metabolism
Fick’s law 19 assessment 215, 216
FIRF see flow scaled impulse hepatic, perfusion CT–PET 225,
residue function 225–6
flow extraction product see FE tumor vascularity and 217,
flow scaled impulse residue 217–21, 219
function (FIRF) 21, 24, 30–1 see also fluorodeoxyglucose
scanning protocols and 36, 37 positron emission
fluorodeoxyglucose positron tomography; tumor
emission tomography vascular–metabolic
(18FDG-PET) relationships
angiogenesis imaging 80, 81 graft-versus-host disease 209, 211
combined with perfusion CT Greece, ancient 2–3
Perfusion see perfusion growth, tumor see tumor growth
computed tomography–
positron emission hardening, beam see beam
tomography hardening effects
lymph node metastases 163 Harvey, William 8–9
pulmonary nodules 115, 119, head and neck cancer 100–6
120–2 imaging 101–2
tumor vascularity and predicting response to therapy
218–21, 219 102–5, 103, 104, 105, 207
[18F]fluoromisonidazole (F-MISO) treatment response assessment
80, 222 200, 200–1
9781842143094-Idx 8/8/07 5:37 PM Page 235
Index 235
hematological malignancies 78 image acquisition protocols 36–7,

hepatic arterial fraction (HAF; 50–5
hepatic perfusion index, HPI) image processing 61–71
35, 62, 183 rectal cancer 132–3, 132–4, 135
calculation 181–3, 182 image segmentation 65–9
dynamic liver scintigraphy 188 different tissue types 68–9, 70, 71
liver metastases 184, 185 specific organs and tumor 66,
pathophysiology of changes 66, 67
189, 190 vascular structures 66, 67–8
slope-ratio methods 176–7 imaging
validation in animal model 41, 42 angiogenesis 78–81, 82–3
hepatic artery 174, 174 history of development 6–8
input to liver 34–5 see also specific imaging
regulation of blood flow 175 modalities
hepatic blood flow 181 impulse residue function (IRF) 20,
arterial (Fa) 181, 182, 183 21, 21–2
portal (Fp) 182, 183 convolution and 22–4, 23
total (FT) 181, 182, 183 flow scaled see flow scaled
hepatic blood volume (HBV) 177 impulse residue function
hepatic metastases see liver India, ancient 2
metastases injection rates, contrast media 55–6
hepatic perfusion index (HPI) see input arterial concentration
hepatic arterial fraction function see arterial
hepatic phosphorylation fraction concentration function, input
index (HPFI) 225–6, 226 integrins 202
hepatocellular carcinoma 141, in vivo videomicroscopy 190–1
141–2 iodine
Hippocrates 2–3 doses injected 55, 58
historical perspective 1–12 enhancement 20, 38
Hounsfield, Sir Godfrey 7–8, 11 perfusion calculations 58, 58–9
Hounsfield units (HU or CT sensitivity, calibration for 114,
number) 20 115, 116
Huang Ti 2 tube voltage and 54, 54
hypoxia, tumor 74, 100, 217 iodine-containing contrast agents
drugs targeting 199–200, 200 see contrast media, X-ray
selective imaging 222 Ito (stellate) cells 175, 191
tumor metabolic response 217,
217–18, 221 Johnson and Wilson model 17–19,
hypoxia inducible factors (HIFs) 18, 29–30
217, 217–18 adiabatic approximation 30–1
arrival time (T0) parameter 34
image(s) hepatic artery fraction 35
frequency/interval 36–7, 51 reproducibility studies 42
number of 51 scanning protocol 37
series, overall length of time 51 validation 40
9781842143094-Idx 8/8/07 5:37 PM Page 236
236 Index
Kety, Seymour 10 validation of CT measurements

Kety–Schmidt method 10, 26 41, 42
kinetic models 26–31 vascularity–metabolism
liver perfusion 177–8 relationship 218, 219, 220
validation 39–41 lung cancer 111–26
see also Johnson and Wilson diagnostic evaluation 117–19, 118
model; two-compartment enhancement thresholds 58
model FDG-PET 115, 120–2
future developments 123–5
laser thermal therapy 206 heterogeneity 122, 123
lenalidomide 74, 78 hypoxia inducible factors 218
Leonardo da Vinci 5 image segmentation 66, 67
liver lymph node metastases 164,
complications of marrow 166, 166
transplantation 209, 211 non-small-cell 112, 120, 124
glucose metabolism, perfusion CT pathology 112
Perfusion–PET 225, 225–6 perfusion thresholds 59
hepatic artery and portal vein potential limitations 122–3
input 34–5 predicting response to therapy
limitations of perfusion CT 191–2 124, 207, 208
perfusion physiology 174, 174–5 presentation 112
quantification of perfusion with radiologic–pathologic
CT 176–83 correlations 119–20
liver metastases 173–93 small-cell 112, 120, 124
factors predicting survival 186, 187 technical aspects 113–17, 114,
in vivo videomicroscopy 190–1 115, 117
novel antiangiogenic agents 202–4 treatment response assessment
occult microscopic 173–4, 185–6 124, 201, 201
other imaging techniques 188–9 vascularity–metabolism
pathogenesis of microcirculatory relationships 218–19,
changes 189–91 219, 221
CT Perfusion for detection see also pulmonary nodules
183–6, 184 lymph node metastases 163–7
renal cancer 156, 157 diagnosis 164, 164–6, 165, 166
vascularity–metabolism growth curves 167, 167
relationships 219, 220 lymph nodes 163–71
liver microcirculation 174–5 enlargement 163
physiopathology of metastatic technical issues 164
changes 189–91 lymphoma 167–71, 169
quantitative measurement 175–83 primary cerebral 94–6
liver scintigraphy, dynamic 188 treatment response monitoring
liver tumors 168–71, 170
primary 141, 141–2
transcatheter arterial magnetic resonance angiography
chemoembolization 206 (MRA) 80
9781842143094-Idx 8/8/07 5:37 PM Page 237
Index 237
magnetic resonance imaging (MRI) Moniz, Egas 6–7

brain tumors 90 Moore, George 10
dynamic contrast-enhanced myelodysplastic syndrome (MDS) 78
(DCE) 79, 81, 189 myeloma, multiple 78
prostate cancer 149
rectal cancer 139 nephrotoxicity, drug 209, 210
mammary adenocarcinoma, Nexavar see sorafenib
R3230AC rat-derived 39–40 noise 38, 54–5
maximum enhancement see peak spatial smoothing 62–4, 63, 64
enhancement no outflow assumption 25–6
maximum slope method 26 validation 39, 43
maximum tolerated dose (MTD)
81, 201–2 oxygen-15-labeled water (H215O)
mean transit time (Tm; MTT) 16, 62 80, 81
average, in region of interest oxygenation, tumor 100–1
64–5, 68–9, 71
central volume principle 19–20 pancreatic tumors 139–41, 140
contrast enhancement protocols parametric maps 61–2, 62, 63, 70–1
56, 56 cerebral tumors 93
liver 177, 181, 182, 183 head and neck tumors 104, 105
reproducibility studies 42–3 liver 183, 184
validation of measurement 40–1 pulmonary nodules 115, 115
MEDI-522 203, 204 rectal cancer 134, 135, 137, 138
meningiomas 91–2, 92, 93, 206 partial volume averaging (PVA) 31–2
metastases Patlak analysis 29
angiogenesis and 74–5 lung cancer 125
brain 96 protocols 49, 53, 57
liver see liver metastases peak (maximum) enhancement
lymph node see lymph node 52–3, 62, 113–14
metastases calculation of perfusion from 58,
renal cancer 156, 157 58–9
micrometastases, detection in liver image segmentation 66, 67, 67–8
173–4, 185–6 pulmonary nodules 114–15, 119
microvessel density (MVD) 202 perfusion see blood flow
lung cancer 120, 124 perfusion computed tomography
lymphoma 168 (CTP) 15–43
prostate cancer 149 basic terms 16–17
rectal cancer 129, 137 history of development 10–12
renal cancer 158, 158 kinetics modeling 26–31
Miles, Ken 11–12 practical issues 36–8
minimum transit time (Tmin; minTT) protocols see protocols,
21, 21–2 perfusion CT Perfusion
contrast enhancement protocols theory 15–36
56, 56 validation 39–43
Mondino de Luzzi 4 without kinetics modeling 25–6
9781842143094-Idx 8/8/07 5:37 PM Page 238
238 Index
perfusion computed perfusion–pathology correlation

tomography–positron emission 152–3, 153, 154, 155
tomography (CT–PET) 48, 126, response to radiation therapy
215–27 153–5, 206
advantages 215, 216 prostate gland, normal
hepatic phosphorylation of 148–50, 149
glucose 225, 225–6, 226 protocols, perfusion CT Perfusion
treatment response assessment 47–59, 48
222–5, 223, 224, 225 choice 57
tumor aggression and 221–2 comparative results 57–9
tumor vascular–metabolic contrast enhancement 55–7
relationships 218–21, 219 example 49
perfusion-weighted map 62, 68, 69 image acquisition 50–5
permeability surface area product pulmonary nodules 114,
see capillary permeability 114–17, 117
surface area product rectal cancer 130–1, 132
phantoms, iodine sensitivity theoretical model used and
115, 116 36–7
physiology PS see capillary permeability
history of development 8–10 surface area product
knowledge in ancient times 1–3 pulmonary metastases,
placental growth factor (PlGF) 75 vascularity–metabolism
platelet-derived growth factor relationships 219, 219
(PDGF) 75, 202 pulmonary nodules 111, 112,
portal vein 174, 174 113–26, 117
input to liver 34–5 diagnostic evaluation 117–19,
regulation of blood flow 175 118
positron emission enhancement/perfusion
tomography (PET) measurements 113–17, 114,
angiogenesis imaging 80, 81 115, 117
combined with perfusion CT fluorodeoxyglucose (FDG) PET
see perfusion computed 115, 119, 120–2
tomography–positron increasing numbers detected
emission tomography 112–13
lymph nodes 163 limitations of perfusion CT
rectal cancer 139 Perfusion 122–3
prediction of response to therapy small 122
206–9 strategy for investigating 121,
hepatic metastases 186, 187 121–2
lung cancer 124, 207, 208 vascularity–metabolism
rectal cancer 136 relationships 219–21, 221
prostate cancer 147–56 see also lung cancer
combined perfusion CT–PET 224
detection and staging 147–52, R3230AC rat-derived mammary
150, 151 adenocarcinoma 39–40
9781842143094-Idx 8/8/07 5:37 PM Page 239
Index 239
radiation dose 49–50 residue function

pulmonary nodules 123 liver perfusion imaging
scanning protocol and 37 178–80, 179
signal-to-noise ratio and 38 see also flow scaled impulse
radiotherapy (RT) residue function; impulse
assessment of complications 209 residue function; tumor
brain tumors 97–8, 98, 99, 206 residue function
predicting response to 102–5, respiration
103, 104, 105, 207 gating 50, 191
prostate cancer 153–5, 206 suspended or quiet 50
response assessment 205–6 respiratory motion 50
tumor oxygenation and 100 correction 37, 116, 192
Radon, Johann 8 kidneys 157
recirculation 35–6 liver 191–2
RECIST criteria 77–8, 197, 198 pulmonary nodules 116–17, 123
rectal cancer 129–39 response assessment, treatment
bevacizumab therapy 77, 138–9, see treatment response
202, 203, 223, 224 assessment
chemoradiotherapy 136–7, 137, Revlimid see lenalidomide
206, 207–8 RMP-7 199, 199, 203
clinical experience 134–9 Röntgen, Wilhelm Conrad 6
combined perfusion CT
Perfusion–PET 223, 224 Santorio, Santorio 9
data and image analysis 132–3, scanning protocols 36–7
132–4, 135 see also image acquisition
patient preparation 130 protocols
technique 130–1, 132 scientific basis 15–43
vascularity–metabolism scintigraphy, dynamic liver 188
relationship 222 segmentation, image see image
see also colorectal cancer segmentation
regions of interest (ROI) 61 signal-to-noise ratio 38
average mean transit time 64–5 single compartment model 49
image segmentation 65–9 skeletal muscle tumors 40–1
liver 174, 176, 183 slices, CT, number and thickness
prostate cancer 152, 153 51–3, 53
pulmonary nodules 114 slope-ratio methods, liver
rectal cancer 134, 135 perfusion imaging 176–7
Renaissance period 4–5 solitary pulmonary nodules see
renal cancer (RCC) 156, 156–8 pulmonary nodules
antiangiogenic therapy 77–8 sorafenib 74, 77–8, 81
metastatic lesions 156, 157 spatial smoothing 62–4, 63, 64
pathologic correlations 158, 159 spiral computed tomography (CT)
reproducibility systems 11–12
animal studies 41, 42 standardized enhancement value
clinical studies 42–3 (SEV) 115
9781842143094-Idx 8/8/07 5:37 PM Page 240
240 Index
standardized perfusion value (SPV) lymphoma 168–71, 170

52, 115 other therapies 206
lung tumors 119, 120–1 phase I trials of novel
renal cancer 158, 158 antivascular agents 201–5
standardized uptake value (SUV) prostate cancer 153–5
115–16, 120–1 radiotherapy 205–6
stellate (Ito) cells 175, 191 rectal cancer 136–9, 137, 138
stroke, ischemic 96–7 tube, X-ray see X-ray tube
SU5416 81 tumor enhancement 58
SU6668 203, 204 calculation of perfusion from 58,
sunitinib (SU11248; Sutent) 74, 58–9
78, 81 peak (maximum) see peak
enhancement
T0 see appearance (arrival) time protocols 55–7
table toggling 52 tumor growth
taxane 200, 203 angiogenesis and 74–5
terms, basic 16–17 nodal metastases and 167, 167
thalidomide 74, 78, 81, 224 tumor perfusion
Thornton, J. B. 1 characteristics 76–7
time–density curve (TDC) comparison with enhancement
arterial 23, 23–4, 35–6 58, 58–9
tumor 21, 22–4, 23, 35–6 measurement 15
tirapazamine 222 see also blood flow
transcatheter arterial tumor progression 74–5
chemoembolization, liver tumor residue function (tumor
tumors 206 TDC) 21, 22–4, 23, 35–6
transfer constant see FE tumor vascular–metabolic
transfer flux of solute through relationships 217–25
capillary endothelium after therapy 222–5
17–19, 18 biomolecular mediators 217,
transfer function, tissue 178, 180 217–18
transrectal ultrasound (TRUS) 147 imaging studies 218–21, 219,
treatment 220, 221
complications, assessing 209 tumor aggression and 221–2
predicting response see tumor vasculature 73
prediction of response to assessing response to treatment
therapy 198–201, 199, 200, 201
treatment response assessment characteristics 76, 76–7
197–211, 203 two-compartment model 27,
brain tumors 99, 198–9, 199, 203 27–9
combined perfusion CT–PET arrival time (T0) parameter 34
222–5, 223, 224, 225 hepatic artery fraction 35
drugs affecting tumor protocols 49
vasculature 198–201, 199, scanning protocol 36, 37
200, 201 validation 39–40
lung cancer 124, 201, 201 see also Patlak analysis
9781842143094-Idx 8/8/07 5:37 PM Page 241
Index 241
ultrasound (US) venous outflow, assumption of no

contrast-enhanced 80, 148 25–6, 39
transrectal (TRUS) 147 Vesalius, Andreas 5–6, 9
see also Doppler ultrasound videomicroscopy, in vivo 190–1
urogenital tract tumors 147–59 volume of distribution see
distribution volume
validation 39–43 VX2 tumor cells 40–1, 42
animal tumor models 39–41, 42
clinical studies 42–3 Weibull functions 180
vascular disrupting agents
(VDAs) 198 X-ray contrast agents see contrast
vascular endothelial growth factor media, X-ray
(VEGF) 75, 202 X-rays
pulmonary nodules 120 discovery 6
therapeutic inhibitors 74, 77 exposure factors 53–5
vascular structures, image X-ray tube
segmentation 66, 67–8 current (mA) 53–5
vatalanib 81 voltage (kVp) 38, 53–5, 54
9781842143094-Idx 8/8/07 5:37 PM Page 242

MTC Oncology

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

MTC Oncology

Încărcat de

Drepturi de autor:

Formate disponibile

9781842143094-FM 8/8/07 12:09 PM Page i

Kenneth A Miles MBBS FRCR MSc MD FCRP

© 2007 Informa UK Ltd

First published in the United Kingdom in 2007 by Informa Healthcare, Telephone

Tel: +44 (0)20 7017 5000

Data available on application

Distributed in North and South America by

Within Continental USA

Distributed in the rest of the world by

Composition by Cepha Imaging Pvt Ltd, Bangalore, India.

To Elizabeth, Matthew, Benjamin, and Samuel …

1 Computed tomography perfusion: a historical perspective 1

2 Scientific basis and validation 15

3 Image acquisition and contrast enhancement 47

5 Angiogenesis, tumor perfusion, and cancer management 73

6 Tumors of the brain, head, and neck 89

7 Perfusion CT applications in lung cancer 111

8 Tumors of the gastro-intestinal tract 129

9 Tumors of the urogenital tract 147

10 CT perfusion of lympth nodes 163

11 CT perfusion of liver metastases and early 173

12 Beyond RECIST: perfusion CT in evaluating 197

13 Perfusion CT–PET: opportunities for combined 215

Daniel Balvay PhD Xiaogang Chen PhD

Ethan J Halpern MD William W Li

Errol Stewart BSC Ian A Yang MBBS PhD FRACP

The extraordinary advances in medical technology over recent years

Professor Miles has undertaken pioneering work in the development

Professor Dame Janet Husband DBE FMedSci PRCR FRCP

relationship to contrast enhancement on CT, and to outline the ability

Perfusion computed tomography (CT) has been made possible by the

ANATOMICAL AND PHYSIOLOGICAL KNOWLEDGE

Many of the medical accomplishments of the great civilizations of

2 Multidetector computed tomography in oncology

Perfusion computed tomography: a historical perspective 3

of wounds. His knowledge of internal organs was thus largely specula-

THE DEVELOPMENT OF MODERN ANATOMY

Anatomy is one of the oldest branches of medicine, and in Western

4 Multidetector computed tomography in oncology

Galen expanded his knowledge partly by experimenting with live ani-

Perfusion computed tomography: a historical perspective 5

Many famous artists of the early Renaissance pursued the study of

6 Multidetector computed tomography in oncology

in medicine and the beginning of modern anatomy. Versalius died in

ANATOMY AND IMAGING

Perfusion computed tomography: a historical perspective 7

in the 1930s, when it was noticed that iodine-containing products

8 Multidetector computed tomography in oncology

PHYSIOLOGY AND THE CIRCULATION

Perfusion computed tomography: a historical perspective 9

10 Multidetector computed tomography in oncology

Figure 1.2 Adolf Fick, who will be

of the Würzburg Physical–Medical Society for July 9, 1870. Interestingly,

PERFUSION COMPUTED TOMOGRAPHY: COMBINING

The anatomical imaging modality of CT and the principles of tracer

Perfusion computed tomography: a historical perspective 11

quantifying vascular physiology comprised attempts to measure cerebral

12 Multidetector computed tomography in oncology

Hospital, Cambridge, UK.19 Calculating perfusion from the maximal rate

Perfusion computed tomography: a historical perspective 13

In this chapter, the theory of computed tomography (CT) tumor perfu-

16 Multidetector computed tomography in oncology

well as the models used to measure them will be further discussed in