Vascular network modeling - Improved parallel implementation on computing cluster

Abstract

In this paper, an improved parallel algorithm of vascular network modeling is presented. The new solution is based on a more decentralized approach. Moreover, in order to accelerate the simulation of vascular growth process both the dynamic load balancing and periodic rebuildings of vascular trees were introduced. The presented method was implemented on a computing cluster with the use of the MPI standard. The experimental results show that the improved algorithm results in better speedup thus making it possible to introduce more physiological details and also perform simulations with a greater number of vessels and cells. Furthermore, the presented approach can bring the model closer to reality where the analogous vascular processes can be also parallel. © 2010 Springer-Verlag Berlin Heidelberg.