XUE Bao-shan, ZHANG Feng-feng, YAN Xiao-jian, et al. Virtual Simulation of Pedicle Screw Placement Based on GPU Acceleration[J]. 2021, 26(4): 125-131.DOI: 10.15938/j.jhust.2021.04.017.
The virtual surgery simulation training system of pedicle screw placement has the problems of high computational difficulty,low visual and tactile refresh rate and poor operability,which leads to poor effect of surgical training. In order to solve this problem,a virtual surgery simulation system of pedicle screw placement based on Graphic Processing Unit ( GPU) acceleration is built in this paper. Firstly,the hybrid data model is used to reconstruct the spine. Secondly,collision detection,force feedback calculation and voxel elimination are used to build the virtual simulation system. Finally,the parallel processing framework of GPU is applied to the system simulation to improve the real-time performance of the system. The results of comparative experiments show that: In the conventional virtual simulation system,the visual refresh rate is about 10 Hz,and the tactile refresh rate is about 50 Hz. In this paper,GPU acceleration technology is used,the visual refresh rate is basically stable at 80Hz,which is increased by 70% ,and the tactile refresh rate is about 600 Hz,which greatly improves the performance,In other words,GPU accelerated virtual surgery system can greatly shorten the computing time of tactile and visual threads,improve the refresh rate,and then improve the operability of the system.