三维不规则窦房结电生理模型的建立及其图形实现

发布时间：2018-11-22 08:16

【摘要】：针对窦房结三维电生理建模复杂、仿真运算量大等问题,提出了一种有效的解决方法。首先利用建模软件建立右心房三维不规则立体模型,再利用网格划分软件剖分为四面体网格,最后根据网格的位置赋予窦房结和心房细胞属性建立电生理模型。数值解算时采用算子分裂法和有限体积法,分别对反应-扩散方程和单纯描述电扩布的微分方程进行了处理,同时利用高性能图形处理单元(GPU)在CUDA(compute unified device architecture)编程环境下实现程序的并行加速。另外,从减少CPU和GPU间的数据交换频率、参量存储方式等多方面对程序加以优化。计算结果表明,所建模型无论是单细胞动作电位还是膜电位的传导均符合正常的电生理特征。通过与串行程序及基于4核8线程CPU编写的共享内存式OpenMP并行程序进行比较,基于GPU的CUDA程序可将运行耗时减少90%以上,CPU和GPU间的数据交换仅占总耗时的3%,且组织模型越大,网格划分越细致,GPU的加速效果愈显著。
[Abstract]:In order to solve the problems of complex 3D electrophysiological modeling and large amount of simulation operation, an effective solution is proposed. First, the 3D irregular three-dimensional model of right atrium was established by modeling software, then the meshing software was used to divide it into tetrahedral meshes. Finally, the electrophysiological model was established according to the location of the grid to endow the sinoatrial node and atrial cell properties. The operator splitting method and the finite volume method are used to deal with the reaction-diffusion equation and the differential equation which simply describes the distribution of electric diffusion. At the same time, the high performance graphics processing unit (GPU) is used to realize the parallel acceleration of the program in CUDA (compute unified device architecture) programming environment. In addition, the program is optimized from the aspects of reducing the frequency of data exchange between CPU and GPU and the way of parameter storage. The results show that both the single cell action potential and the membrane potential conduction accord with the normal electrophysiological characteristics. Compared with serial program and shared memory OpenMP parallel program based on 4-core 8-thread CPU, the CUDA program based on GPU can reduce the running time by more than 90%, and the data exchange between CPU and GPU is only 3% of the total time consuming. The larger the organizational model and the more detailed the mesh division, the more significant the acceleration effect of GPU is.
【作者单位】：西安交通大学电气工程学院;
【基金】：国家自然科学基金资助项目(81271661) 陕西省自然科学基金资助项目(2016JM8069)
【分类号】：R541;TP391.41

【相似文献】