An algorithm for rotor tracking in atrial fibrillation using graph search-based periodic peak detection.

Rotors are rotating electrical waves that may sustain atrial fibrillation (AF); thereby providing therapeutic targets for catheter ablation. We propose a method for identifying rotors from circular catheter recordings of bipolar intracardiac electrograms (EGM) during AF. We use dominant frequency-based periodicity detection along with a graph search algorithm to identify the most dominant periodic activations or peaks of interest in each bipolar EGM recorded by a multipolar circular catheter. We then track the activations across catheter bipoles to determine whether they conform to the rotational pattern of a rotor. The performance of the proposed method is tested on simulated bipolar EGM arrays containing rotor activation corrupted by noise and complex aperiodic signal features. The method is shown to perform with high accuracy (up to 98% sensitivity and 100% specificity) in detecting simulated rotors and may serve to guide rotor ablation in patients with AF.