Weakly-Coupled Oscillators with Long-Distance Correlation as a Model of Human Atrial Fibrillation

Atrial Fibrillation (AF) is a chronic, progressive, and heterogeneous disease, which exhibits irregular and chaotic electrical activity in atria. The exact mechanisms of AF have remained elusive. The complexity of the atrial signals, lack of repeatability, and non-local nature of AF make it very difficult to characterize its spatiotemporal organization. This paper presents a set of data-processing tools to find and evaluate the intracardiac signals in AF based on synchronization theory. Specifically, a graph-theoretical algorithm is developed to prune spurious intracardiac spikes. The processing pipeline was applied to 10 datasets recorded during ablation procedures in patients with paroxysmal or persistent AF. AF cases are classified into three main types. Type I (n=3) is defined as the presence of a single driver that dominates both atria and is consistent with the mother rotor hypothesis. The majority of the cases had more than one driver and were classified as types II (n=4) and III (n=3). The drivers, i.e., interacting organized areas, in type II have significantly different peak frequencies and are only weakly coupled. The drivers in type III have close frequencies, are moderately interacting, and show evidence of transient intermittent phase-locking (intermittency). In addition, a long-distance correlation between channels from the left and right atria was observed. The theory of synchronization is a useful conceptual framework to process and analyze AF and can provide mechanistic insight into the mechanisms of AF and the effect of various interventions.