Clinical Anatomy 26:486492 (2013)

ORIGINAL COMMUNICATION

A Comparison of Hepatic Segmental Anatomy as Revealed by Cross-Sections and MPR CT Imaging

XUE-JING LIU,1,2 { JIAN-FEI ZHANG,3 { HONG-JIN SUI,3* SHENG-BO YU,3 JIN GONG,3 JIE LIU,4 LE-BIN WU,1* CHENG LIU,1 JIAN BAI,5 AND BING-YI SHI5
Shandong Medical Imaging Research Institute, Jinan, China Department of Radiology, Xuan Wu Hospital, Capital University of Medical Sciences, Beijing, China 3 Department of anatomy, Dalian Medical University, Dalian, China 4 Dalian Hoffen Bio-Technique Co. Ltd., Dalian, China 5 Organ Transplantation Center, The Second Afliated Hospital of Chinese PLA General Hospital, Beijing, China
2 1

To compare the areas of human liver horizontal sections with computed tomography (CT) images and to evaluate whether the subsegments determined by CT are consistent with the actual anatomy. Six human cadaver livers were made into horizontal slices with multislice spiral CT three-dimensional (3D) reconstruction was used during infusion process. Each liver segment was displayed using different color, and 3D images of the portal and hepatic vein were reconstructed. Each segmental area was measured on CT-reconstructed images, which were compared with the actual area on the sections of the same liver. The measurements were performed at four key levels namely: (1) the three hepatic veins, (2) the left, and (3) the right branch of portal vein (PV), and (4) caudal to the bifurcation of the PV. By dividing the sum of these areas by the total area of the liver, the authors got the percentage of the incorrectly determined subsegmental areas. In addition to these percentage values, the maximum distances of the radiologically determined intersegmental boundaries from the true anatomic boundaries were measured. On the four key levels, an average of 28.64 6 10.26% of the hepatic area of CT images was attributed to an incorrect segment. The mean-maximum error between articial segments on images and actual anatomical segments was 3.81 6 1.37 cm. The correlation between radiological segmenting method and actual anatomy was poor. The hepatic segments being divided strictly according to the branching point of the PV could be more informative during liver segmental resection. Clin. Anat. 26:486492, 2013. V 2013 Wiley Periodicals, Inc.
C

Key words: computed tomography; hepatic and portal veins; segmental anatomy; multicolor infusion

{ Xue-Jing Liu and Jian-Fei Zhang have contributed equally to this work.

Medical Imaging Research Institute, Jinan 250021, China. E-mail: cjr.wulebin@vip.163.com Received 13 December 2009; Revised 4 March 2012; Accepted 1 April 2012 Published online 10 May 2013 in Wiley (wileyonlinelibrary.com). DOI 10.1002/ca.22095 Online Library

Contract grant sponsor: National Natural Science Foundation of China; Contract grand number: 31071052 *Correspondence to: Prof. Hong-Jin Sui, Ph.D. Department of Anatomy, Dalian Medical University, Liaoning 116044, China. E-mail: suihj@hotmail.com and Prof. Le-Bin Wu, Shandong

C 2013 V

Wiley Periodicals, Inc.

Comparative Study of Hepatic Segmental with Cross-Section and CT Hepatic resection has been growing over the last decade. Delineation of the portal venous segments and subsegments in the human liver has become increasingly important to radiologists, particularly because of the growing need for accurate preoperative localization of focal intrahepatic lesions and the need for living donor liver transplantation (LDLT). Different authors have proposed the division of the liver into two hemilivers, into four segments based on the Goldsmith and Woodburne (Goldsmith et al., 1957) denition or into eight subsegments based on the Couinaud (1957) denition, which is now considered the international standard (Soyer, 1993; Gazelle et al., 1994). Understanding the segmental anatomy of the liver has been a vital component of this expansion and has stimulated technical developments. In hepatic surgery, medical imaging is used to detect and localize hepatic lesions and their relationship to vascular structures (Soyer et al., 1996; Togo et al., 1998; Bassignani et al., 2001; Atasoy et al., 2004; Torabi et al., 2008). In particular, the portal vein (PV) denes the livers functional anatomy into several discrete segments. With the development of hepatic surgery, this segmental classication was partly authenticated but it also proved to be very inaccurate (Fischer et al., 2002). Many liver cancer patients also have cirrhosis, which decreases the contrast of computed tomography (CT) images of the hepatic PV and hepatic vein (HV) when compared with normal (Kang, 2002). Some studies could not reconstruct satisfactory three-dimensional (3D) images for the liver segments determined. As the hepatic portal venous system is of great importance for liver surgery and interventional procedures, any congenital and acquired anomalies will make it difcult to dene liver segments accurately (Gallego et al., 2002). Furthermore, the intersegmental boundaries are complex and undulating interfaces rather than simple, at planes (Zhang et al., 2008). Therefore, the traditional liver segmental classication used in CT imaging does not accord with the real anatomical segments. However, the 2D CT images of all patients clearly show the PV and HV and the segments of liver were divided by straight lines on the 2D images before. But the invisible intersegmental vessels and complex undulating interfaces were not displayed clearly through the 2D CT images. Therefore, the authors adopted the international advanced plastination technique, combined with multicolor infused and postprocessing CT techniques. The authors compared multiplanar reconstruction 2D CT images with plastinated horizontal sections to determine the veracity and practicability of Couinauds classication and traditional liver segmental classication, used in radiological imaging. Furthermore, the authors wanted to nd a better method to solve the difculty in dening clinical liver segments.

487

Republic of China) were examined. They were surgically removed at autopsy with the preservation of the hilar vascular pedicle and the inferior vena cava (IVC). After the identication of the origins of the PV, the proper hepatic artery (HA) and bile duct, the superior margin of the IVC was prepared by double suturing. The inferior margins of the IVC, PV, and HA were then cannulated with polyethylene tubing and connected to latex tubing for infusion. The livers were ushed gently with water and any identied leaks were sewn up. These livers were scanned by multislice spiral CT (MSCT) scanner (Philips AG, Amsterdam, The Netherlands) with 100 ml of air injected into the PV (Fig. 1) before cannulation of every isolated hepatic segment. The CT scanning was performed with the following parameters: section thickness of 1 mm, pitch of 1, reconstruction interval of 0.5 mm, 120 kV, and 200 mAs. The 3D image of PV was reconstructed with shaded surface display technique (Kimberly et al., 2001) during this period. These images were then used as the basis for liver segment cannulation and infusion. During scanning, the livers were put into a mold made by polyurethane foam, this mold made the liver keep the shape and position just like they were in the abdomen cavity.

Cannulating and Infusing the Subsegments

The PV was dissected from the visceral surface of the liver to expose the left branch of the portal vein (LPV) and the right branch of the portal vein (RPV) completely. Then the LPV and RPV were opened longitudinally along the branches and polyethylene tubing was cannulated into every isolated hepatic segment and ligated securely. The PV branch of every hepatic subsegment was infused with a different color silicone (R1, China) and 2.5% Ultravist 300 (Schering AG, Berlin-Wedding, Germany). During the process of multicolor infusion of portal venous branches, CT scans were performed to obtain the data of subsegmental portal venous branches (Fig. 2). After the injections of other segments had been completed, segment I and the trunk of the PV were injected. Then 60 ml air was injected into the HV, and these livers were scanned again to allow their 3D reconstruction. Evacuating the air from the liver and infusing the HV with white silicone was the last step of infusion. All these procedures were based on Couinauds and Bismuths classication systems. They were also based on the 3D reconstructed CT image of the PV (with air infused). The HA was then infused with 10% formalin and the livers were then placed in 10% formalin for 3 days to x the tissues.

Sawing the Liver into Slices and Plastinating Them

The xed livers were frozen at 258C. Before sawing the liver, MSCT scanning was performed. The CT scanning was performed with the following parameters: section thickness of 1 mm, pitch of 1, reconstruction interval of 0.5 mm, 120 kV, and 200 mAs. The 3D image of PV was reconstructed with the 3D

MATERIALS AND METHODS

CT Scanning
In this study, six-donated fresh human liver specimens (Dalian Hoffen Bio-Technique, Dalian, Peoples

Fig. 1. 3D reconstruction images of PV with air injection.

Fig. 2. The CT images of liver when II and VIII subsegmental branches of PV were infused. a: Axial CT image. b: VR reconstructed image. c: MIP reconstructed image.

Fig. 3.

Level of three hepatic vein trunks. a: Liver slice. b: MPR CT image.

Comparative Study of Hepatic Segmental with Cross-Section and CT software volume rendering (VR) at this period. The frozen liver was then embedded in polyurethane foam and sawed into 3 mm thick horizontal slices using a band saw. About 1 mm was lost in the sawing process. Each specimen produced 3245 horizontal slices. The two surfaces of each liver slice were washed gently with water and all tissue fragments were removed. Photographs and data of these liver slices were gained. The liver slices were then plastinated (Gao et al., 2006). This process enabled us to make the slices easy to preserve and restudy when needed.

489

Comparisons and Measurements

The authors observed the six liver horizontal slices to determine whether there were intersegmental hepatic veins between two adjacent segments. The authors made multiplanar reformation CT images (MPR-CT) to conform to the segments of infused liver slices. The traditional CT-determined subsegments were compared with the authentic anatomic subsegments obtained from direct investigation of the six liver specimens and the authors measured differences between the CT images and actual anatomical segments. The measurement was performed on the CT sections at four key levels: at the three HV trunks, at the LPV, at the RPV, and caudal to the bifurcation of the PV. The total area of the liver was measured on CT sections. For each segment, the authors measured the area as delineated on CT sections that did not correspond to the actual subsegment on these horizontal slices. By dividing the sum of these areas by the total area of the liver, the authors derived the percentage of incorrectly determined subsegmental areas. In addition to these percentage values, the maximum distances of the radiologically determined intersegmental boundaries from the true anatomic boundaries were measured.

Subsegment VIII was not between subsegments V and VI as described by Fasel (1998), but between subsegments V and IV. Subsegment VII was found to extend caudally behind subsegment VI. From these image comparisons, we found that the segments of liver judged by radiologists on the four key levels of CT images (Figs. 3b, 4b, 5b, and 6b) had many differences from their real segmental ascriptions (Figs. 3a, 4a, 5a, and 6a). There were incorrect judgments on segments at all four key levels of the livers (Table 1). An average of 28.64 6 10.26% of the hepatic area on axial CT scans was attributed to the incorrect segment. The maximum distance between articial segments on images and actual anatomical segments ranged from 1.93 to 6.68 cm with a mean of 3.81 6 1.37 cm (Table 2).

DISCUSSION
Procedures for delineating segmental and subsegmental liver anatomy on CT and magnetic resonance (MR) images have been the subject of several studies during the past 20 years (Soyer et al., 1996). Essentially, these procedures are based on the concept of three vertical planes that divide the liver into four segments, and of transverse boundaries that further subdivide the segments into two subsegments each. With the development of the techniques, increasing numbers of problems appeared in the eld of hepatic surgery and radiologic imaging of liver (Strunk et al., 2003). Fasel et al. (1998) found that 19.1, 50.20, 52.9, and 15.4% of the hepatic areas visualized on axial CT scans were attributed to the incorrect subsegment at the four key levels listed above, respectively. Moreover, the maximum deviation of the subsegmental boundaries, as determined radiologically on axial CT scans, was 40 mm (Fasel et al., 1998). Although in our study, the authors found that this incorrect subsegment allocation was not as high, at the four key levels the incorrect measures reached areas of 23.40, 24.63, 36.57, and 29.96%. A mean of 28.64 6 10.26% of the hepatic area on axial CT scans was attributed to the incorrect segment. The range of incorrect areas in individual specimens ranged from 15.58 to 36.05%. Our study found that the incorrect segment denitions at the cranial and caudal regions on axial CT scans were lower than at central regions near the right and left branches of the PV. These data are shown in absolute values in Table 1. The mean-maximum distance between articial segments on MPR CT images and actual anatomical segments was 3.81 6 1.37 cm. Nelson et al. and Soyer et al. also found that there were 31 and 22% incorrect segmental areas dened on 2D CT liver images (Soyer et al., 1994). Some previous research studies concerning segmental anatomy of the liver adopted the corrosion cast technique, but the covering of small branches made it difcult to distinguish the pattern of the PV branches. Thus, there was not a relevant report about segmental infusion. This study obtained the CT 3D image of PV before intubation. It allowed us to determine which segment the vessels went into. By

RESULT
The correlation was poor between articial segments on CT images and actual anatomical segments (Figs. 36). Figures 3a, 4a, 5a, and 6a show the horizontal slices with the subsegments infused. Different colors on these pictures show the different subsegments and the borders between colors are the real boundaries. These proved to be complex and undulating interfaces rather than simple at planes and varied between specimens. Figures 3b, 4b, 5b, and 6b are MPR-CT images of the same four key levels in which the subsegments delineated radiologically and anatomically are compared. At the level of three hepatic vein trunks (Fig. 3a), the caudal subsegment III can be identied even at this cranial level. At the level of the LPV (Fig. 4a), parts of subsegment II can be seen. Subsegment V can be identied at the level of the RPV (Fig. 5a). Subsegment V is between subsegments VII and VIII, and part of subsegment II can also be identied. At the level caudal to the PV bifurcation (Fig. 6a) subsegments III, IV, V, and VI could be found, but part of subsegments VII and VIII could also be seen.

490

Liu et al.

Fig. 4.

Level of left branch of portal vein. a: Liver slice. b: MPR CT image.

Fig. 5.

Level of right branch of portal vein. a: Liver slice. b: MPR CT image.

Fig. 6. Level of caudal to the portal vein bifurcation. a: Liver slice. b: MPR CT image. (The numbers 18 on the pictures denoted the segment I*VIII of human liver.)

Comparative Study of Hepatic Segmental with Cross-Section and CT

TABLE 1. The Percent of Incorrect Judgment with Segments at the Four Key Levels (%)a Levels 1 2 3 4 Average
a

491

Liver 1 25.80 29.17 26.11 25.50 26.64

Liver 2 15.64 13.87 17.55 15.25 15.58

Liver 3 23.93 23.26 51.07 45.93 36.05

Liver 4 33.88 23.50 48.49 31.36 34.31

Liver 5 16.23 25.30 35.67 32.31 27.38

Liver 6 24.89 32.67 40.53 29.42 31.88

Average 23.40 24.63 36.57 29.96 28.64

Level 1: hepatic vein trunks; level 2: left branch of portal vein; level 3: right branch of portal vein; level 4: caudal to portal vein bifurcation.

TABLE 2. The Distance Between Articial Segments on MPR CT Images and Actual Anatomical Segments at Four Key Levels (cm)a Levels 1 2 3 4 Average
a

Liver 1 3.02 4.22 3.18 4.22 3.66

Liver 2 2.03 3.20 3.50 2.34 2.77

Liver 3 2.60 5.57 6.60 6.68 5.36

Liver 4 5.24 5.04 4.90 3.24 4.61

Liver 5 2.50 3.67 4.86 4.13 3.79

Liver 6 1.93 2.89 3.78 2.04 2.66

Average 2.89 4.10 4.47 3.78 3.81

Level 1: hepatic vein trunks; level 2: left branch of portal vein; level 3: right branch of portal vein; level 4: caudal to portal vein bifurcation.

using the knowledge obtained with the CT scanning and observing carefully during dissection and cannulation, the origin of PV branches could be accurately judged. This method solved the problem that segmental hepatic vessels cannot be infused. At the same time, the MSCT images of the liver segments could be compared with the images of CT 3D image of PV gained at the beginning. Furthermore, the air could be removed easily and had no adverse effects for the next operation. This research also showed the subsegments with different colors on the surface. This demonstration is clearer, more intuitive to the observer and more anatomically accurate. The CT ndings thus conrm that the assumption of a at, transverse separation between the cranial and caudal subsegments is far from anatomic reality. Therefore, our conclusion supports the results of Fasel et al. In our research, the authors made MPR-CT images to conform to liver transverse slices. Because the 16-slice and 64-slice spiral CT have isotropic advantages, these liver specimens could be scanned by spiral CT before being sawed into transverse section. After liver sections were made, these CT images could be reconstructed according to the position of the liver section. Then these reconstructed 2D images of the livers were compared with the liver transverse sections. This comparative research was more accurate than ever. Thus, the hepatic segments being divided strictly according to the branching point of the PV could be more reliable during liver segmental resection.

and Dalian and First Afliated Hospital of Dalian Medical University supported this study.

REFERENCES
Addis KA, Hopper KD, Iyriboz TA, Kasales CJ, Liu Y, Wise SW. 2001. Optimization of shaded surface display for CT angiography. Acad Radiol 8:976981. Atasoy C, Akyar S. 2004. Multidetector CT: Contributions in liver imaging. Eur J Radiol 52:217. Bassignani MJ, Fulcher AS, Szucs RA, Chong WK, Prasad UR, Marcos A. 2001. Use of imaging for living donor liver transplantation. Radiographics 21:3952. Couinaud C. 1957. Le foie: etudes anatomiques et chirurgicales. Paris: Masson. Fasel JH, Selle D, Evertsz CJ, Terrier F, Peitgen HO, Gailloud P. 1998. Segmental anatomy of the liver: Poor correlation with CT. Radiology 206:151156. Fischer L, Cardenas C, Thorn M, Benner A, Grenacher L, Vetter M, Lehnert T, Klar E, Meinzer HP, Lamade W. 2002. Limits of Couinauds liver segment classication: a quantitative computerbased three-dimensional analysis. J Comput Assist Tomogr 26:962967. Gallego C, Velasco M, Marcuello P, Tejedor D, De Campo L, Friera A. 2002. Congenital and acquired anomalies of the portal venous system. Radiographics 22:141159. Gao H, Liu J, Yu S, Sui H. 2006. A new polyester technique for sheet plastination. J Int Soc Plastination 21:710. Gazelle GS, Lee MJ, Mueller PR. 1994. Cholangiographic segmental anatomy of the liver. Radiographics 14:10051013. Goldsmith NA, Woodburne RT. 1957. The surgical anatomy pertaining to liver resection. Surg Gynecol Obstet 105:310318. Kang HK, Jeong YY, Choi JH, Chung TW, Seo JJ, Kim JK, Yoon W, Park JG. 2002. Three-dimensional multidetector row CT portal venography in the evaluation of portosystemic collateral vessels in liver cirrhosis. Radiographics 22:10531061. Soyer P. 1993. Segmental anatomy of the liver: Utility of a nomenclature accepted worldwide. AJR Am J Roentgenol 161:572573. Soyer P, Bluemke DA, Bliss DF, Woodhouse CE, Fishman EK. 1994. Surgical segmental anatomy of the liver: demonstration with

ACKNOWLEDGMENTS
The authors thank Timothy Seekings for proofreading the manuscript. Dalian Hoffen Bio-technique

492

Liu et al.
Togo S, Shimada H, Kanemura E, Shizawa R, Endo I, Takahashi T, Tanaka K. 1998. Usefulness of three-dimensional computed tomography for anatomic liver resection: Sub-subsegmentectomy. Surgery 123:7378. Torabi M, Hosseinzadeh K, Federle MP. 2008. CT of nonneoplastic hepatic vascular and perfusion disorders. Radiographics 28:19671982. Zhang JF, Yu SB, Liu J, Liu XJ, Sui HJ. 2008. Boundaries between subsegments IVa and IVb in the human liver. Clin Anat 21:439446.

spiral CT during arterial portography and multiplanar reconstruction. AJR Am J Roentgenol 163:99103. Soyer P, Heath D, Bluemke DA, Choti MA, Kuhlman JE, Reichle R, Fishman EK. 1996. Three-dimensional helical CT of intrahepatic venous structures: comparison of three rendering techniques. J Comput Assist Tomogr 20:122127. Strunk H, Stuckmann G, Textor J, Willinek W. 2003. Limitations and pitfalls of Couinauds segmentation of the liver in transaxial imaging. Eur Radiol 13:24722482.