Research On Virtual Endoscopy Technology Based On Volume Rendering

Compared with the conventional medical endoscopic techniques, non-invasive, economical, no perforation and no infection risk, high sensitivity, it can detect the site or organ that the conventional endoscope can not check, they are the advantages of virtual endoscopic techniques. It regards the rule-based grid scalar data field and the medical volume data with a certain resolution as the virtual endoscopy's processing objects, and its key problems is how to three-dimensional reconstruct these entered medical data, that is, how to reconstruct the various data from CT, MRI and other medical imaging equipment, fast, clear and complete reproduced in front of users.At present, the three-dimensional reconstruction method of medical image, there are two main categories: indirect rendering method and direct rendering method. At present, the applied research of direct drawing method in a virtual endoscopic technique is also an urgent and an important task that need to solve. Direct volume rendering algorithm has a large amount of calculation, and it has a slow rendering speed relatively, which greatly limits its application, so to explore how to improve the volume rendering speed and quality problems, for improving the theory of three-dimensional reconstruction, expanding the application of volume rendering has important theoretical and practical significance.In this paper, I propose a new parallel computing and real-time rendering method of virtual endoscopy scene based on GPU, using this method, the algorithm decreases the amount of data collection of image memory in the condition that doesn't influence the accuracy of the 3D-image display. And by using the effective abilities of parallel computation and multi-data process, it can calculate the complex ray-generation and rendering clips in the volume rendering.