Numerical evaluation of cardiac mechanical markers as estimators of the electrical activation time

Abstract

Recent advances in the development of non-invasive cardiac imaging technologies have made it possible to measure longitudinal and circumferential strains at a high spatial resolution also at intramural level. Local mechanical activation times derived from these strains can be used as non-invasive estimates of electrical activation, in order to determine e.g. the origin of premature ectopic beats during focal arrhythmias or the pathway of reentrant circuits. The aim of this work is to assess the reliability of mechanical activation time markers derived from longitudinal and circumferential strains, denoted by AT ell and AT ecc , respectively, by means of three-dimensional cardiac electromechanical simulations. These markers are compared against the electrical activation time (AT v), computed from the action potential waveform, and the reference mechanical activation markers derived from the active tension and fiber strain waveforms, denoted by AT ta and AT ef f , respectively. Our numerical simulations are based on a strongly coupled electromechanical model, including Bidomain representation of the cardiac tissue, mechano-electric (i.e. stretch-activated channels) and geometric feedbacks, transversely isotropic strain energy function for the description of passive mechanics and detailed membrane and excitation-contraction coupling models. The results have shown that, during endocardial and epicardial ectopic stimulations, all the mechanical markers considered are highly correlated with AT v , exhibiting correlation coefficients larger than 0.8. However, during multiple endocardial stimulations, mimicking the ventricular sinus rhythm, the mechanical markers are less correlated with the electrical activation time, because of the more complex resulting excitation sequence. Moroever, the inspection of the endocardial and epicardial isochrones has shown that the AT ell and AT ecc mechanical activation sequences reproduce only some qualitative features of the electrical activation sequence, such as the areas of early and late activation, but in some cases they might yield wrong excitation sources and significantly different isochrones patterns.

Enabling forward uncertainty quantification and sensitivity analysis in cardiac electrophysiology by reduced order modeling and machine learning

“…So far, only few papers have addressed a systematic uncertainty quantification (UQ) and propagation in cardiac electrophysiology models, despite the very rapid growth of cardiac modeling and the need to deal with multiple scenarios, towards model personalization. For instance, the role of infarct scar dimensions, border zone repolarization properties and anisotropy in the origin and maintenance of cardiac reentry has been considered in Reference 75; the evaluation of cardiac mechanical markers (such as longitudinal and circumferential strains) to estimate the electrical activation times has been addressed in Reference 76. See, for example References 2,77, for a review of the possible sources of uncertainty in these models at different spatial scales.…”