Abstract
Improving the performance of executing virtual reality applications is a challenge for researchers. In this paper we present a parallel virtual reality application of the human femur and hip joints on a cluster of processors. The core of parallel processing is the use of HCL (High- performance Communication Library), MPICH, and WinMPI (a personal computer version of MPICH) to perform message passing between the scaleable processors. After computer tomography (CT) images are acquired from scanning devices, segmentation is performed to find the contours of each cross section image. Finite element models are generated from automesh. We have implemented a data-parallel algorithm for the surface and volume rendering to reconstruct 3D image models, which can be cut at any sections or blocks to show their internal structures. Our parallel rendering algorithm has shown the performance improvement on virtual reality applications. Applying our prototype system to low-cost personal computers makes it more acceptable for clinical applications. They can be used to help doctors in dealing with the clinical and pathological problems of patients. In the future we would like to develop virtual reality applications of surgery simulation that can assist surgeons and medical students to plan and practice operations more effectively.