Atrial Fibrillation Mapping Using Phase-Aligned Spectral Filtering for Decomposing Spatiotemporal Dynamics

Atrial fibrillation (AF) ablation is an important treatment strategy, potentially capable of curing AF. However, there is a variable degree of success of the persistent AF ablation procedure. Therefore, it is necessary to improve the success rate of persistent AF ablation. To improve the success rate of persistent AF ablation, the investigators proposed a novel analytical approach to AF mapping. Analysis of intracardiac electrograms (EGMs) is the key component of AF mapping. In the past, several approaches to atrial EGM analysis have been developed, aiming to help guide AF ablation beyond pulmonary vein isolation (PVI). However, none (except PVI) have been established to guide the AF catheter ablation procedure. The most widely used approach for AF ablation is an empiric, anatomic AF ablation approach (PVI). Previously tested (and ultimately failed) methods included complex fractionated atrial electrogram (CFAÉ), dominant frequency (DF) mapping, activation (FIRM) mapping, and fibrosis-detected-by-CMR-mapping AF ablation. A novel solution to the known AF mapping problem is necessary to improve AF ablation outcomes. Successful completion of the proposed project will suggest a path to improve the success rate of persistent AF ablation, which is crucially necessary for further progress in the cardiac electrophysiology field. The investigators proposed a novel analytical approach. The investigators will apply phase-aligned spectral filtering to AF atrial EGM mapping data to identify spatially structured dynamic components and thus identify AF ablation targets. Spectral filtering is an efficient dimension reduction for high-dimensional time series, which has not been applied to AF mapping data analysis. Such an approach assumes that the observed spatiotemporal data (AF EGM mapping data) represent superimposed lower-rank smooth oscillations and movements from a dynamic generative system (AF ablation target) mixed with higher-rank random noises. Separating the signals from noises is essential for us to locate and understand these lower-rank dynamic systems. It could be that such a lower-rank dynamic system has multiple independent components corresponding to different trends or functionalities of the system. The investigators propose a novel framework for identifying lower-rank dynamics and their components embedded in a high-dimensional spatiotemporal system (AF EGM map). It is based on a known statistical approach of structural decomposition and phase-aligned construction in the frequency domain, which has not been applied to AF EGM analysis.