An inverse framework for estimating cardiac electrophysiological activity from medical image sequence

Noninvasive estimation of cardiac electrophysiological activity (myocardial electrical activity) traditionally used potential measurements on the human body surface, such as body surface potentials (BSPs, pl.) or electrocardiograms (ECGs, pl.). But because of non-unique mapping between myocardial electrophysiological activity and sparse body potential measurements, the performances of previous efforts have not been well accepted. Furthermore, few progresses have been made in noninvasive estimation of myocardial electrophysiological activity recently. In this thesis, a different aspect to understand myocardial electrophysiological activity is established for the first time: since the mechanical activity of the myocardium is mainly driven by myocardial electrophysiological activity, patient-specific myocardial kinematic measures should, indirectly, reflect the propagation of cardiac transmembrane potential (TMP). In the implementation, a meshfree particle representation of myocardial volume and its fiber structure is developed, upon which the spatiotemporal electrophysiological activity and electromechanical coupling can be properly simulated using the FitzHugh-Nagumo (FHN) model and electromechanical coupling model respectively. Coupling the physiological modeling of myocardial electro-mechanical behavior on the meshfree particle representation, my proposed inverse approach recovers myocardial electrophysiological activity from medical image sequences through regularization or Kalman filter with favor results