Estimation of Purkinje Activation from ECG: An Intermittent Left Bundle Branch Block Study

Modelling the cardiac electrophysiology (EP) can help understand pathologies and predict the response to therapies such as cardiac resynchronization. To this end, estimating patient-specific model parameters is crucial. In the case of patients with bundle branch blocks (BBB), part of the Purkinje system is often affected. The aim of this work is to estimate the activation of the right and left Purkinje systems from standard non-invasive techniques: magnetic resonance imaging (MRI) and 12-lead electrocardiogram (ECG). As it is difficult to differentiate the contribution of the Purkinje system, this work relies on a particular intermittent left BBB (LBBB) case where both LBBB and absence of LBBB (ALBBB) were recorded on different 12-lead ECGs. First, an efficient forward EP model is proposed by coupling a Mitchell-Schaeffer cardiac model with a current dipole formulation that simulates the ECG. We used the Covariance Matrix Adaptation Evolution Strategy (CMAES) algorithm to optimize the 3 parameters by minimizing the error with the real ECG. The estimation of conduction velocity (CV) parameters for LBBB and ALBBB shows a good agreement on the myocardial CV (0.39m/s for ABBB, 0.40m/s for LBBB), while the estimation of the left Purkinje CV seems to identify the pathology (1.32m/s for ALBBB, 0.49m/s for LBBB). Finally, the plots of the simulated 12-lead ECGs together with the ground truth ECGs indicate similar shapes.