Modeling Atrial Activation Sequences in the Rabbit Heart in Macroscopic and Microscopic Dimensions

Severe atrial conduction disturbances are the most common arrhythmias in the elderly population. The heart undergoes a remodeling process with normal aging. Among all age related changes, increased fibrosis, which alters cardiac structure and the excitation spread at a microscopic scale, may play an important role in the genesis of atrial flutter (AFL) and fibrillation (AF). Research on the atrial electrical activation sequence comprising microstructural aspects of fibrosis is therefore of great interest. This thesis is focused on the modeling and simulation of the electrical excitation spread in the right atrial isthmus which is seen as a critical region for AFL and AF. The computer models are based on signals from electrophysiological in vitro experiments with Rabbit hearts and on histological tissue samples provided by cooperating research groups. Simulating cardiac excitation spread with realistic ion kinetics and in microscopic details is a very challenging task due to the tremendous computational effort involved. In this work, state of the art kinetic models for atrial cells and meshing techniques were employed to build up tissue models in the range of some centimeters in 2D and for some aspects of 3D. In addition to that, a new method to accelerate the simulations using graphic-hardware has been implemented. The computer models were employed to study and elucidate some key questions of electrophysiological mechanisms. It could be shown that superfusion technique with oxygenated solution is an adequate experimental protocol for studying the spread of depolarization in thin-walled tissue preparations. The structure-related computer models developed here allowed to investigate the impact of macro- and microstructures on depolarization waveforms at the surface of the tissue. The microscopic model showed in detail the directional dependence of electrogram morphology in regions representing different types of microstructures. These findings can be used to discriminate different classes of fibrosis. The implementation of a numerical solver to simulate cardiac cellular dynamics on graphics processing units (GPUs) outperformed the parallel implementations of the same algorithm on a small cluster of computers. The results of this work can be seen as a promising step towards whole heart simulations performed at relatively low costs compared to high performance computing facilities.