Quantitative Analysis of Rotor Distribution in Atrial Fibrillation via Noninvasive Electrocardiographic Imaging

Roland Sanford1, Jonathan Chrispin2, Saman Nazarian3, Linwei Wang1
1Rochester Institute of Technology, 2John Hopkins University, 3University of Pennsylvania


Abstract

Atrial Fibrillation (AF) is the most common cardiac arrhythmia found in people of all ages. The irregular electrical signals that cause the arrhythmia are thought to be sustained by abnormalities in the heart’s tissue. These waves can be described as rotors, which appear to orbit a central region called the anchor. Current clinical solutions involve invasively mapping the epicardial and endocardial potentials to locate arrhythmia-causing tissue of either class. Once identified, the problematic tissue regions become targets of high-energy catheter ablation. Unfortunately, invasive procedures are associated with numerous issues that may contribute to suboptimal ablation outcomes, including limited accuracy and procedural risk. Noninvasive Electrocardiographic Imaging (ECGI) is a computational process that estimates the epicardial potentials from the patient’s geometry and body surface potentials (BSPs). In previous studies, ECGI has shown promise in localizing arrhythmia sources in AF, as verified by the results of ablation and invasive mapping data. However, detailed ECGI-based quantitative characterization of the spatiotemporal organization of AF rotors has not been presented. We present a novel study to quantify the spatiotemporal rotor distribution in the context of AF based on tracking phase singularity points (PSPs) in ECGI results. Patient-specific geometric models and recorded body-surface potentials (BSPs) serve as inputs to the computational pipeline. Second-order Tikhonov regularization is used to compute estimates of the epicardial potentials based. Dominant frequency analysis, phase mapping, and phase singularity identification serve as tools that, in conjunction with clinical insight, identify potential ablation sites. To date, thirteen total patients have been analyzed using our computational pipeline. In ECGI computed results, at each time instant when PSPs are identified, an average of (4.78 +/- 2.28) rotors were found. The majority of rotors, through stable in location, were transient in time.