The Research Of Optimization And Parallelization For Cardiac Computational Model Based On Reaction Diffusion System

Arrhythmia is a group of cardiovascular diseases with high morbidity and mortality.In order to study the inner mechanism of arrhythmia,the research of cardiac electrophysiological model has made great progress.At present,on the micro level of cardiac,a new research filed has just emerged which to be aimed at reveal the connection between the microscopic expression with the macroscopic pathological features through the new type of ion channel,such as TRPM ion channel model.Meanwhile,to build a multiscale heart model which including the level from micro to macroscopic,reaction diffusion system is the most mainstream way in this field.Therefore,building a multiscale cardiac electrophysiological model based on the reaction diffusion system to describe the whole progress which from the expression of ion channel to the generation of electrical excitation to the spread of excitement,has a certain theoretical significance and practical value.This paper will transplant and design a one-dimensional cardiac model of Takeuchi HL-1 atrial cell based on TRPM ion channel,and based this onedimensional model,a two-dimensional cardiac tissue model will be built with the aid of reaction diffusion system.In addition,this paper will study the performance optimization and parallelization acceleration technology to improve the established model.Firstly,we put forward a decoupling idea of the reaction diffusion system,then proposed a performance optimization strategy for the one-dimensional model and presented a parallel acceleration method for the two-dimensional model.The experimental results show that the performance of these model has been improved.The main contributions of this paper are as follows:1)A two-dimensional cardiac electrophysiological model based on the reaction diffusion system.This paper transplanted and designed the Takeuchi HL-1 atrial myocyte model,which can clearly explain the influence of TRPM ion channel expression on the macroscopic pathological changes.On this basis,this paper designed and implemented a two-dimensional cardiac electrophysiological model based on the reaction diffusion system,which enable us to reveal the pathogenesis of cardiac arrhythmia in a more comprehensive way.2)A cache strategy based on the shared date of ion channels for the onedimensional Takeuchi HL-1 atrial myocyte model.This paper analyzed the model principle and calculation modules,and proposed a cache strategy based on the shared date of ion channels,which can avoid the reduplicative read of data.The contrast experiments show that this optimization strategy can effectively eliminate the costs for calculating repeated data,to improve the computational efficiency.3)A parallel acceleration method based on GPU for the two-dimensional cardiac electrophysiological model based on the reaction diffusion system.On the one hand,in order to easily employ parallelism,this paper simplified the reaction diffusion equation to the decoupling state,that is to say,the reaction module and the diffusion module can be solved independently after the simplification.On the other hand,because the accurate propagation of 2-D model has a very high demand in computing performance,this paper proposed a parallel acceleration method based on GPU general computing.It is found that there is a high degree of parallelism in the diffusion module between cells in two dimensional tissues.In this paper,GPU is a good environment for parallel acceleration,and the experiments show that GPU's acceleration ratio can reach to 68 times or more.