Shannon Energy Application for Detection of ECG R-Peak Using Bandpass Filter and Stockwell Transform Methods

Abstract

Shannon energy-based algorithm has been implemented in peak detection method of various physiological signals including electrocardiogram, which is used to enhance significant peaks for accurate peak detection. Two significant methods of R-peak detection that apply Shannon energy are identified. However, direct comparison cannot be made due to the differences in database used, number of beat analysed, frequency range selected, and signal processing technique applied. This paper aimed to properly evaluate the performance of Shannon energy-based algorithms for R-peak detection on two methods of bandpass filter and Stockwell transform. Simple enveloping technique using moving average filter is proposed, and a threshold is set to localize R-peak at a selected frequency range of 7-15 Hz. Performance of both methods were then evaluated using all 48 data from MIT-BIH Arrhythmia database. Result showed that both methods are equivalently useful in reducing P and T waves interference and produced similar output of Shannon energy envelope. However, Shannon energy application on bandpass filter offered 99.71% sensitivity, 99.80% positive predictivity and 99.52% accuracy, slightly better than that of the Stockwell transform method that only produced 99.65% sensitivity, 99.68% positive predictivity and 99.33% accuracy.