Inverse ECG Modeling with Spatiotemporal Regularization for the Characterization of Myocardial Infarctions

Abstract

Myocardial infarctions (MIs) pose a significant risk to human health. Accurate identification and characterization of MI's are essential for the effective medical treatment. Traditional methods such as the standard 12-lead ECG identify MIs with the electrocardiogram (ECG) recorded on the body surface, which consider little about anatomical details of the human body. These methods are limited in the ability to map back the actual electrical activities of the heart and further characterize MIs. Inverse ECG (iECG) methods were proposed to trace the distribution of electric potentials on the heart surface and characterize MIs. However, these methods do not account for the spatiotemporal behaviors of the potential distributions, because the electric potentials are distributed in the complex geometry and varying dynamically over time. In this paper, a novel iECG model with spatiotemporal regularization is developed to image and characterize MIs. We solve the iECG problem with the method of spatiotemporal regularization and reconstruct electric potentials on the heart surface. Furthermore, we group the estimated heart potentials into healthy and infarct clusters with a wavelet-clustering method. Experimental results show that the proposed method effectively solves the iECG problem and better characterizes MIs compared with existing methods.