Detection of atrial fibrillation in electrocardiogram signals using tunable-Q wavelet transform

Abstract

This work introduces the application of wavelet entropy for detection of episodes of atrial fibrillation (AF), from electrocardiogram (ECG) recording. AF is a common cardiac arrhythmia which is often asymptomatic and shows very brief episodes. Early detection of AF will improve the treatment of it and will reduce the risk of death and strokes in patients. In the present work after getting the RR segment we apply tunable Q wavelet transform (TQWT) and then calculate the wavelet entropy of that RR segment. We repeat this process for different values of Q. In the presence of AF the complexity of signal increases which results in higher value of wavelet entropy. Thus on the basis of value of wavelet entropy we decide the presence or absence of AF in a beat. This method provides a good discriminant ability of 93.44% which have been compared with previous work. This method is capable of detecting very brief episodes of AF, which are hardly of few beats in length. This method can be implemented in ECG monitoring system, it can help clinicians in early detection of AF and to gain sufficient knowledge of causes that results in arrhythmia.