Integrated image processing and electro-anatomical mapping to assess arrhythmogenicity of the infarct border zone

Ischemic heart disease and specifically myocardial infarction (MI) causes regional damage (scarring) to the heart and is responsible for up to 75-80% of sudden cardiac death (SCD). The common therapy to reduce SCD is an implantable cardioverter defibrillators (ICD). In primary prevention, the main quantitative criterion for ICD implantation remains left ventricular ejection fraction (LVEF). This parameter has poor accuracy to predict arrhythmia risk and is a measure of pump function therefore not reflecting the underlying arrhythmia mechanisms. Methodologies that probe arrhythmia mechanisms more directly have the potential to improve risk prediction in patients after myocardial infarction.

In this project novel methodology was developed and combined with existing approaches to assess different mechanism underlying arrhythmias after MI, based on cardiac imaging and electrophysiological mapping and with a focus on assessing regional properties in the scar border zone. All were developed in a porcine model of MI, based on a transient occlusion of the left-anterior descending artery, providing an established validation setting. The first novelty is the detailed analysis of structural heterogeneity of the infarct scar. Late gadolinium enhanced in-vivo MR-imaging of MI was done. Manual segmentation was performed to extract voxels of the left ventricle. Left ventricle infarct voxels were identified as voxels with signal intensity exceeding 50% of the signal intensity range. Intertwined fibrosis and intact myocyte voxels were automatically annotated using novel computational homology software. The abundance and circumference of intertwined regions provide a measure of structural infarct heterogeneity. Secondly, a novel method to determine the activation and repolarization time, based on the electric field, was introduced and validated against monophasic action potential recordings. This methodology allowed spatially dense mapping of activation and repolarization times of the endocardial surface. Initial evaluation using a porcine animal model of myocardial infarction revealed increased spatial and temporal heterogeneity in the border zone. Furthermore, we observed that adrenergic stimulation further amplified heterogeneity. Therefore, these two measures were integrated with previously described nuclear imaging in a novel pipeline. Regional differences in innervation and perfusion were analyzed using 11C-epinephrine and 13N-ammonia PET-MR imaging. In a porcine model of myocardial infarction resulting from left anterior descending artery occlusion/reperfusion it was shown that the innervation/perfusion mismatch regions correspond with regions of increased repolarization heterogeneity.

Three novel methodologies were developed to capture individual mechanism underlying ventricular arrhythmias. These methods were developed and evaluated using a porcine animal model of infarction and suggest more detailed insight in the culprits of arrhythmias. Furthermore, all employed techniques have the potential to be translated to the clinical setting with no or minimal modification.