Calibration of a fully coupled electromechanical meshless computational model of the heart with experimental data

Most electromechanical cardiac models available in the literature are based on the Finite Element Method (FEM). Despite FEM’s high accuracy, its performance is intrinsically dependent on the geometrical mesh quality onto which equations are discretized and solved. The construction and personalization of meshes representing the high complexity and inter-subject variability of anatomical structures such as the heart are often challenging and require tedious manual adjustments. Meshless methods are an interesting alternative to overcome meshing difficulties. The majority of meshless methods have been applied to simulate fluid dynamics problems but rarely in cardiac tissue electromechanics due to the lack of advanced numerical techniques to correct instabilities near the boundary conditions. In this paper, we propose a fully coupled model of cardiac electromechanics based on Total Lagrangian Smoothed Particle Hydrodynamics (TL-SPH). A zero-dimensional (0D) circulation model of hemodynamics was coupled to the 3D electromechanical model, linking ventricular volume changes along the cardiac cycle with pressures exerted on the endocardial surface of the heart. For calibration purposes, the proposed TL-SPH model was personalized to minimally invasive in vivo pressure measurements and in vivo magnetic resonance imaging (MRI) data of the left ventricle (LV) of a canine heart. The parameters of the electromechanics and hemodynamics models were estimated to fit LV volumetric and displacement data from MRI and available pressure signals in two phases: first, passive parameters based on diastolic data minimizing point-based differences between simulations and measurements; and, subsequently without any pressure information, active contraction and hemodynamics parameters based on volumetric and pressure data over the entire cardiac cycle. The TL-SPH model was able to reproduce pressure, volumetric and morphological indices from experimental measurements with good accuracy: end systolic / diastolic errors of 0.29 / 0.46 % and 1.82 / 4.18 % for maximal volume and pressure measurements, respectively. As this was achieved without the burden of mesh generation, the obtained results support the use of meshless methods as a valid alternative to standard FEM for cardiac electromechanical modeling, especially to assimilate point-based cardiac medical data.