Abstract

ABSTRACT Computational models of cardiac function are increasingly considered in industry for designing medical device therapies, and, in the clinic for diagnosis and therapy planning to tailor patient-specific therapies. A fundamental concern hampering a broader adoption is the lack of evidence of a close correspondence between the physiology of a virtual heart and physical reality. Creating such evidenceremains challenging as biophysically detailed virtual hearts are characterized by high dimensional parameter vectors that must be identified from limited low dimensional, noisy and uncertain observations. Further, even for carefully calibrated models, their ability to provide patient-specific predictions of the cardiac response to therapies based on their mechanistic nature is assumed, but not proven. Finally, generating detailed mechanistic models requires complex computationally costly workflows, requiring operators with significant skill levels. This raises concerns regarding scalability to applications to larger variable patient cohorts, the validity of insights produced by error prone workflows, as well as the reproducibility of in silico studies. These concerns render advanced industrial and clinical applications often unviable from both a regulatory as well as an economic perspective.Here, we report on methodological advances addressing these issues. Specifically, we present methods for i) the automated generation of anatomically and structurally accurate models of whole heart and torso from medical images, with suitable reference frames to support automation of parameter sweeps; ii) full physics real-time enabled whole heart electrophysiology simulations and associated electrograms and ECGs; iii) a calibration technique for whole heart electrophysiology using noninvasive ECG measurements; iv) model calibration techniques for cardiac device therapies replicating device measurement applicable for optimizing device designs in industry and for personalized therapy planning in the clinic; and v) a computational approach for guiding ventricular tachycardia ablation therapies based on electrogram and ECG matching.