High performance simulation for brain tumours growth using parabolic equation on heterogeneous parallel computer system

摘要

Brain tumour is one of the prevalent cancers in the world that lead to death. Based on the present knowledge of the properties of gliomas, mathematical models has been developed by researchers to quantify the proliferation and invasion dynamics of glioma within anatomically accurate heterogeneous brain tissue. This paper focuses on the implementation of parallel algorithm for the simulation of brain tumours growth using one dimensional parabolic equation, design on a distributed parallel computer system. The numerical finite-difference method is focused on a design of a platform for discretizing the parabolic equations. The result of finite difference approximation using explicit, Crank-Nicolson and fully implicit methods will be presented graphically. The implementation of parallel algorithm based on parallel computing system is used to capture the growth of brain tumour. Parallel Virtual Machine (PVM) is emphasized as communication platform in parallel computer systems. The software system functions to enable a collection of heterogeneous computers to be used as synchronize and flexible concurrent computational resource. The parallel performance measurement will be analyzed from the aspect of speedup, efficiency, effectiveness and temporal performance.