CRPC-TR99806-S August 1999 Title: Simulating Electrical Action Potential Propagation in the Heart with Parallel Computation Author: Christa Erwin Submitted November 1999 Abstract: One of the central questions in the field of cardiac dynamics is the mechanism of the decay of ventricular tachycardia, characterized by spiral waves of electrical activity, to fibrillation, characterized by "incoherent" electrical wave behavior, resulting in failure of the heart's pumping action. This problem is being chipped away at from various perspectives by various groups in physiology, mathematics, physics and engineering. Numerical simulation, coupled with experimental feedback, is a necessary and powerful tool in this area, as in many other growing areas of computational science. The focus of this summer project was to explore the effect of geometry and fiber architecture of the left ventricle on electrical wave dynamics. This question was movivated by earlier munerical experiments performed in a rectangular slab model of the ventricle, which showed that the rotating anistropy inherent in cardiac tissue could lead to wave instablility. This rotating anistropy is pictured in Figure 1 (from [4] and [5]). Our goal was to construct a minimally realistic geometrical model of the left ventricle and to simulate electrical wave propagation in this three-dimensional model. Dissection results reveal a nested layered geometry for the left ventricle, where a single macroscopic muscle fiber bundle starting at the basal plane outside the midwall (toward the epicardium) traverses down toward the apex on an outer surface, and at some point before reaching the apex, changes direction, traverses back along an inner surface reinserting at the basal plane inside the midwall (toward the endocardium) [6]. The simple fact that electrical propagation along a fiber is several times faster than perpendicular to it suggests that this nested architecture can be important in the propagaion of electrical waves in the heart. In particular, transmural propagation can still be achieved in a nested geometry even when propagation perpendicular to the fiber is weakened or cut off. A geometrical model of the left ventricle has been developed by Sima Setayeshgar in the Applied Mathematics Department at the California Institute of Technology [1], based on previous works [2], [3]. In this model, the fiber surfaces are given by nested cones, and the fiber trajectories by geodesics on these surfaces, consistent with experimental observation. Figure 2 shows the fiber trajectories on one fiber surface. The objective of this summer project was to investigate the effect of the realistic (a) nested conical geometry and (b) fiber architecture of the left ventricle on action potential propagation, using the above model. In the process, the project provided an introduction to the following: - Numerical methods for scientific computing, in particular, finite difference solutions to nonlinear partial differential equations - Parallel computing using MPI - HTML form submission and CGI programming as a front-end for running parallel code on Caltech's Boewulf machine. ------------------------------------------------------------------------------ Christa Erwin CRPC Summer Research Program in Parallel Computing for Undergraduate Women California Institute of Technology