Robust approach for evaluating periodicity in human atrial fibrillation bipolar electrograms

Atrial fibrillation (AF) is a common arrhythmia which can arise from periodic electrical sources. Bipolar electrograms (EGMs) recorded in the atrium are used to localize these sources. However, currently used signal processing approaches, including Fourier transform, do not accurately detect periodicity in bipolar EGMs owing to their complex morphology. This study introduces a novel variant of the periodicity transform (PT) which incorporates a correlation-based signal projection in order to improve detection of periodicities during AF. For validation, synthetic periodic signals added to bipolar atrial EGMs from patients without periodicity were used. The effect of noise and quasi-periodicity was further evaluated. When compared to PT, the proposed method's accuracy in detecting periodicity, at an average signal-to-noise ratio of 0 dB, improved by up to 36%.