Pipeline to Build and Test Robust 3D T1 Mapping-Based Heart Models for EP Interventions: Preliminary Results

Abstract

Computational models are powerful tools in electrophysiology (EP), helping us understand and predict arrhythmia associated with heart attack (i.e., myocardial infarction), a major cause of sudden cardiac death. Our broad aim is to combine novel scar imaging methods with fast computational models to enable accurate predictions of electrical wave propagation, and then to test these models in preclinical frameworks prior to clinical translation. In this work we used n = 3 swine with chronic infarct, which underwent MR followed by conventional x-ray guided electro-anatomical EP mapping. For scar imaging, we employed our T1-mapping MR method based on multi-contrast late enhancement (MCLE) at 1 × 1 mm in-plane resolution and 5 mm slice thickness. Next, we used the MCLE images as input to a fuzzy-logic algorithm and segmented the infarcted area into two zones: infarct core IC (dense fibrosis) and grey-zone, GZ (i.e., arrhythmia substrate). We further built 3D heart models from the stack of segmented 2D MCLE images, integrating tissue zones (healthy, IC and GZ) into detailed tetrahedral heart meshes (~1.5 mm element size). Finally, we investigated the accuracy of model predictions by comparing measured maps of activation times (i.e., depolarization times) with simulated maps obtained by employing a macroscopic formalism and reaction-diffusion equations. We obtained an acceptable small mean absolute error between the simulated and measured depolarization times (~12 ms, in average). Future work will focus on refining MR imaging resolution and use the models to guide ablation procedures.