在人体腿部的虚拟仿真研究中,建立有限元模型并进行生物力学特性分析是一种有效的方法.但由于有限元模型数据量大,解算时间长,并且难以与既有系统融合,因此不适合在实时的手术培训和手术预演中应用.为了提高虚拟手术仿真系统进行实际作业的能力,笔者提出了以BP神经网络模型来代替有限元模型,实现实时的生物力学响应.并结合已有的医疗机器人辅助接骨虚拟现实仿真手术系统,构建了系统实验平台.实验结果证明,人体腿部的BP神经网络模型能够完全满足手术仿真所需的实时性要求.
It is an effective method to establish the finite element model and analyze its biomedical characteristic in the research of the virtual simulation on human leg. However, because of the large volume of data, long solving time, and the difficulty to integrate with the existing system, the finite element model is unsuitable to use in the surgery training and practice. To enhance the operational capacity of the virtual surgery simulation system, a BP neural network model is presented to replace the finite element model, which can realize real - time biomechanical response. Combined with the existing virtual simulation system of Robot assisted orthopedic system, the experimental platform is constructed. The experimental result shows that the BP neuram network of human leg could fully satisfy the requirement of real- time simulation.