Efficient estimation of cardiac conductivities: a Proper Generalized Decomposition approach

While the potential groundbreaking role of mathematical modeling in electrophysiology has been demonstrated for therapies like cardiac resynchronization or catheter ablation, its extensive use in clinics is prevented by the need of an accurate customized conductivity identification. Data assimilation techniques are, in general, used to identify parameters that cannot be measured directly, especially in patient-specific settings. Yet, they may be computationally demanding. This conflicts with the clinical timelines and volumes of patients to analyze. In this paper, we adopt a model reduction technique, developed by F. Chinesta and his collaborators in the last 15 years, called Proper Generalized Decomposition (PGD), to accelerate the estimation of the cardiac conductivities required in the modeling of the cardiac electrical dynamics. Specifically, we resort to the Monodomain Inverse Conductivity Problem (MICP) deeply investigated in the literature in the last five years. We provide a significant proof of concept that PGD is a breakthrough in solving the MICP within reasonable timelines. As PGD relies on the offline/online paradigm and does not need any preliminary knowledge of the high-fidelity solution, we show that the PGD online phase estimates the conductivities in real-time for both two-dimensional and three-dimensional cases, including a patient-specific ventricle.