Modeling Electromechanical Activity in the Heart: an Application to Cardiac Resynchronization Therapy.

Personalized patient-specific electromechanical models of the heart can be a huge asset for clinicians to improve therapy selection in multiple cardiac diseases. However, personalization of 3D models is still challenging due to the complexity of the system itself (coupling of electrics, hemodynamics and mechanics), the high number of parameters to be tuned and the computational demand. The Bestel-Clement-Sorine model was used to describe the cardiac electromechanical activity. The heart is described as a passive hyperelastic material accounting for elasticity and friction in the cardiac extracellular matrix surrounding the fibers. Electrical stimulation is derived from an Eikonal activation map computed beforehand and is coupled to the active contraction part, which accounts for the active stress along cardiac fibres and elasticity between sarcomeres and Z-discs. To personalize our model, we used the Covariance Matrix Adaptation Evolution Strategy (CMA-ES), a stochastic method for real-parameter optimization of non-linear functions. The parameters are sampled according to a multivariate normal distribution. The covariance matrix of this distribution is iteratively updated, such as the new set of parameters minimizes the cost function. As a result of this step, we can obtain pressure and volume curves as well as compute electromechanical dyssynchrony parameters well within physiological ranges. However, a few indices of interest are still outside good and expected physiological values. To account for this problem, a further sensitivity analysis was carried out to investigate the impact of action potential duration (APD) increase in mechanical output parameters used in cardiac resynchronization therapy (CRT). Particularly in those markers that were difficult to estimate by only using CMA-ES.