Sunteți pe pagina 1din 1

Direct Volumetric Rendering Based on Point Primitives in OpenGL

André Luiz Miranda da Rosa, Ilana de Almeida Souza, Adilson Yuuji Hira, Marcelo Knörich Zuffo
EPUSP - Escola Politécnica da USP – Engenharia Elétrica – LSI (Laboratório de Sistemas Integráveis)
São Paulo/SP, Brasil; {amiranda, iasouza, ayhira, mkzuffo}@lsi.usp.br

To make diagnosis, doctors many times analyze images becomes a point with the color calculated by the R, G and
slices acquired from, for example, Computed B values related to this pixel position in the corresponding
Tomography, and try to mentally reconstruct the resultant images. Therefore, it is possible to visualize a cloud of
tri-dimensional scene. The acquired information strongly points placed one to the side of the others, through a
depends on the experience and training of the medical distance that allows each voxel to assume the size of one
specialist and its knowledge of the current studied object pixel in the users screen (Figure 1). An interaction model
[Meinzer et al 1991]. Therefore, there’s a necessity to with mouse and keyboard was also created, allowing to
offer a mechanism that facilitates the three-dimensional the user to translate, rotate and scale the model in any
visualization of medical examinations, so volumetric direction.
renderization appeared as a method to project
multidimensional data set in a bi-dimensional
visualization plan to gain better knowledge of the
structure contained in the volumetric data.

Nowadays, there are many graphical boards in the market


that perform the volumetric data renderization by means
of their hardware. As they’re special hardware to
accelerate the volumetric renderization, they offer better
speed than software based renderization algorithms, but
they have the disadvantage to have a higher cost.
Therefore, the use of the software based renderization
system described here is more appropriate for many Figure 1: Visible Human Volumetric Renderization
situations where we don’t want to acquire more expensive
component and only want to use components that come As this direct visualization algorithm only plot the pixels
with personal computers as accelerating boards that in 3D, transforming them into voxels, it becomes
support the OpenGL libraries, becoming very useful for extremely simple. However, he does not offer a good
the majority of the population. renderization speed (in the case of processing all voxels,
without eliminating hidden voxels), with 15
The aim of this paper is to describe a volumetric data frames/second for low-resolution images (183 x 148
renderization method based on point primitives that is pixels and 150 slices) and 3 frames/second for average
simpler than other existing algorithm and that offers a resolution (438 x 354 pixels and 150 slices), depending on
good performance without the use of special graphical the processor and the video board used.
boards. The volumetric data can be acquired, for example,
from archives in TIFF, JPEG, BMP, GIF or DICOM Currently, we are optimizing this algorithm so that hidden
(Digital Imaging Communication in Medicine - a standard voxels would not be processed and it would provide faster
wide used for medical examinations images as Computed renderization and will allow its use in real time
Tomography) formats. Particularly, we chose the applications. It is also being added a pixel size control
segmented visible human data set [Souza 2004] because it (used to represent each voxel) in relation to the user point
is commonly used by the scientific community. of view (so that volume zoom can occur) preventing the
undesirable effect of empty spaces originated by the
Our intention is to integrate this visualization technique to approach of the volume to the user point of view.
the Collaborative Environment for Second Medical
Opinion developed by the Telemedicine Group in the References
LSI/EPUSP that allows the remote collaboration between Meinzer , H., Meetz, K., Scheppelmann, D., Engelmann, U.,
users through the Internet using Chat, Videoconference Baur, H. J. (1991) "The Heidelberg Ray Tracing Model",
and Bi-dimensional Collaborative Blackboard to allow German Cancer Research Center, Heidelberg. IEEE Computer
images diagnosis at distance. Graphics & Applications.

Souza, I.A., Sanches-Jr, C., Binatto, M.B., Lopes, T.T., Zuffo,


This algorithm was implemented using JAVA language M.K. (2004). Direct Volume Rendering of the Visible Human
and graphical library LWJGL (Lightweight Java Game Dataset on a Distributed Multiprojection Immersive
Library). The considered algorithm creates a model in Environment, Anais do VII Symposium on Virtual Reality
OpenGL, through primitive of points, where each voxel (SVR’04), São Paulo, SP, October, 183-194.

S-ar putea să vă placă și