Predicting Electrical Cardioversion outcome from surface ECG recordings through Wavelet Sample Entropy

Electrical Cardioversion (ECV) is the most effective alternative to revert persistent atrial fibrillation (PAF) back to normal sinus rhythm (NSR). It would be clinically useful to predict NSR maintenance likelihood after ECV before it is attempted because of the high atrial fibrillation (AF) risk of recurrence and because of the potential secondary effects of ECV. In previous studies, other parameters have been deeply analyzed, but their results were very different and, consequently, inconclusive. Thereby, this work presents a new non-invasive predictor of ECV outcome before it is attempted. This method is based on the suitable combination of the Wavelet Transform (WT) and sample entropy (SampEn), which is a non-linear regularity index, and is called Wavelet Sample Entropy (WSE). Results indicated that 17 out of 21 (80.95%) ECVs relapsing to AF and 12 out of 14 (85.71%) ECVs resulting in NSR after the first month were correctly discriminated. The recordings that relapsed to AF presented higher SampEn values (0.0320 plusmn 0.0053) than the recordings in NSR (0.0271 plusmn 0.0045). In addition, the five patients in which NSR was not restored in any case, presented the highest SampEn values (0.0350 plusmn 0.0028).