A Research To The Therr-dimensional Visualization System Of Medical Image

Medical visualization is not only a very important visual field in modern visualization, but also is the most popular visualization research fields today. Direct volume rendering is widely used in many fields for it can effectively display three-dimensional structure as well as structure information, contributing to the development and advancement of the medical visualization.With the rapid increase in computer hardware, especially the progress in programmable graphics’ processing ability, a series of volume rendering algorithms based on GPU have been produced,and they have showed the internal structure of three-dimensional data field more quickly and effectively. Conversely, higher requirements in research also encourage more advanced methods to meet demands, and illumination can help improve rendering results, enriching the content of the rendering images. Now there are several platforms integrating with many volume rendering algorithms, which could satisfy most of rendering requirements. Global illumination can be able to greatly increase rendering quality by simulating light refraction and light reflection. However,it is rarely achieved because of its high computational cost. Another research on brain nerve fibers data also plays an important role in neural medicine, it is necessary to expand the existing volume rendering platforms to render brain nerve fiber.In order to enhance the rendering effect of medical volume data and achieve the function of rendering brain nerve fibers, this article does some research and explorations on those problems.Taking advantage of the graphics hardware superiority, we implement the ray casting algorithm based on GPU and other volume rendering algorithms with supporting interactions with images.Then, this system has achieved global illumination, so that we can express more details about volume. Moreover, we have also accomplished the function of rendering brain nerve fibers.(1) Implementations of ray casting based on GPU and other algorithms. In order to improve the rendering efficiency, we have implemented ray casting based on GPU using GLSL. Moreover,we have integrated most volume rendering methods to satisfy the actual demands. Users can perform some interactions on images, which is helpful to follow volume information.(2) Global illumination. By simulating light reflection and light refraction, we calculate all energy particles in volume space, and also blend into the light fusion, thus the result can showglobal illumination. Each time to adjust the transfer function, we will recalculate particle energy and render a new results finally. An apparent contrast shows that images with global illumination show more depth information, strengthen the brightness, promote the rendering quality, and obtain better results.(3) Rendering the brain nerve fibers. In order to integrate the function of rendering brain nerve fibers, this article renders the fiber according to the voxels’ coordinate attributes. Fibers data is a set of nodes sequence, which contains voxels’ coordinate attributes. Just connecting adjacent nodes to fit the entire fibers, we can visual the brain fibers at the end. There exists tracking errors in the tracking process; however, we remove the part of fibers with lower reliability based on fibers mean squared error, achieving the rendering result with higher degree of confidence eventually.(4) Implementation of Medical visualization system. This article utilizes software development kits such as OpenGL, GLSL to achieve ray casting rendering, and uses Qt to establish the system. Our system contains most relative volume rendering algorithms, which achieves global illumination function and be able to render brain fibers. What’s more, it is easy to be maintained and upgraded, because it is based on independent models.