The Research On Ray Casting For Visualization Of Anatomical Structures Based On The GPU

Three-dimensional Visualization technologies of medical images can provide intuitive,distinct and realistic volume data of anatomical structures. They can increase the efficiency of diagnosis effectively, and can help doctors confirm the diagnosis, as well as guide the surgery. Therefore, they have become the hotspot in medical fields in recent years. In this thesis, based on the research on the technologies of visualization, and under grant of the National Natural Science Fund Project "An orthognathic surgical planning and simulation based on 3D statistical deformation and finite element hybrid model", we focus on the subject of Volumeric rendering of 3D medical images. The main contents of this thesis are listed as follows:1. The principle of volumetric rendering is analyzed by the pre-integral synthesis, and Optical models,Ⅲumination models and the acceleration technique of volumetric rendering have been studied systematically. We introduce three classical algorithms of the volume rendering, and simulate the Splatting algorithm.2. To improve low sampling and poor sampling efficiency in the traditional ray casting algorithm, an improved ray casting algorithm based on GPU is proposed. The algorithm combines Blinn-Phong illumination model with a new sample synthesis function and early opacity termination condition, which is implemented by vertex programs and fragment programs in Cg language. The experimental results show that the algorithm can not only enhance the visualization effect of traditional ray-casting algorithm, but also accelerate the speed of volumetric rendering. 3. To improve the low efficiency and poor visualization results in the traditional ray casting algorithm based on GPU, an Octree Encoding based ray casting algorithm is proposed in CUDA, First, in this algorithm, the volume data is structured by Octree, which can eliminate a part of voxels that have no contribution to the reconstructed image efficiently. Secondly, the voxels are bound to 3D texture, and the threads are distributed according to the size of Volume data, parallel computing is used when finding out the intersection of each of the light and the volume data. Finally, the ray casting algorithm is implemented in CUDA kernel. The experimental results show that the proposed algorithm can not only enhance the visualization effect of GPU based traditional ray-casting algorithm, but also accelerate the speed of volumetric rendering.The algorithm proposed in this thesis provide a good visualization tool in helping doctors adjutant detection,pre-operative analysis and intra-operative monitor.