Doctor of Philosophy

dissertationComputer models of the heart can provide a greater understanding of the mechanisms of arrhythmias as well as tools to develop treatment strategies, yet adoption of computation modeling in biomedical sciences lags behind that in other fields. One of the reasons for the slow adoption of computational models in medicine is the lack of robust validation studies. The goal of this dissertation was to develop and apply validation approaches to two types of computer heart modeling pipelines, electrocardiographic (ECG) forward simulation and a defibrillation simulation, by comparing measured and predicted potential fields. Previous validation studies have shown that ECG forward simulations produce greater error than expected, which could be caused by insufficient sampling of the cardiac sources. Various sampling strategies over the atrial region were tested to determine the effect of spatial sampling on the forward simulation. Including atrial samples reduced the error in predicted body-surface potentials, with some strategies more effective than others. These findings could help improve measurement protocols when validating the ECG forward simulation and provide more insight into ways to improve ECG imaging techniques. Simulations of defibrillators have previously been developed to provide patient-specific guidance for improving treatment of fatal arrhythmias. To demonstrate the accuracy of one of these simulations, torso-tank experiments and clinical studies were used to record the potential fields generated by a defibrillator and compared to predicted values. Measurements within the torso-tank, including within the myocardium, and body-surface recordings from patients agreed with corresponding simulated potentials. Predicted defibrillation thresholds (DFTs) also agreed with values observed clinically and experimentally. The simulation's accuracy in predicting potential fields and DFTs supports its use in guiding defibrillation treatment