Comparative study of methods for atrial fibrillation cycle length estimation in fractionated electrograms

Complex fractionated atrial electrograms (CFAEs) are electrograms (EGMs) characterized by a high variability both in amplitude and waveforms, making the estimation of their cycle length (CL) a difficult task to perform. The CL is a widely employed parameter for the characterization of the electrical activity within the atria, thus serving to guide catheter ablation, one of the most effective cardiac procedures for the treatment of AF nowadays. The present study aims to compare the performance of three different methods estimating the CL from a set of 50 recordings with CFAEs previously annotated by two expert physicians. The first algorithm is based on an adaptive amplitude threshold, while the second method performs the detection of the highest activations and a later detection of peaks within intervals longer than 1.5 times the median cycle length of the EGM. Finally, we introduce a novel hybrid method which performs a main amplitude detection, which is facilitated by the equalization of high and low amplitude activations in CFAEs according to a fractionation parameter, and a subsequent search for lower activations within intervals longer than the median CL, decreasing the amplitude threshold proportionally to the intervals' length. Our method outperformed the other analyzed detectors, which confirms that the method proposed is a more accurate estimator of atrial CL for CFAEs.