Spurious Rotor Detection During Atrial Fibrillation: Phase Singularities in Fact Reflect Blurred Conduction Block

Rotors during atrial fibrillation (AF) have been implicated as potential targets for AF ablation. Analysis in the phase domain to detect phase singularities (PS) may however blur the underlying wave front activation pattern, and obscure lines of conduction block. We investigated the coincidence of PS trajectories and lines of conduction block to assess the disruptive effect of phase reconstruction methods on the underlying propagation pattern and rotor detection accuracy. High-density unipolar electrograms were recorded in 20 patients with AF during cardiac surgery, using an electrode array (25x25 mm) placed on the left posterior and right atrial wall. Activation maps were constructed using probabilistic local deflection detection; phase maps were reconstructed using sinusoidal recomposition or frequency-based filtering, followed by phase computation using Hilbert transform or time-delay embedding. Conduction block was defined as local conduction velocity < 10 cm/s between adjacent electrode activations. In 38 recordings (10 seconds), 138 sustained PS trajectories were detected (lifespan > 1 atrial cycle), using sinusoidal recomposition and Hilbert transform, out of which 130 (94%) were consistently within 1 electrode distance (1.5 mm) of a line of conduction block. In contrast, concurrent points in the mapping array were consistently further away from a line of block than the detected PS points (median difference +2.1 mm, p << 0.001). Other approaches to reconstruct phase maps yielded comparable results. The far majority of rotors detected in phase maps are co-located with a line of block in activation maps, not with a rotating wave front, showing that rotor detection by phase singularity tracking is a non-specific method in the presence of conduction block. Our results may explain why ablation at the core of a rotor can sometimes be successful.