A Novel Compound Method for R-peak Detection in Electrocardiograms

Nasrin Bastani¹ Raheleh Kafieh² Vahid Bastani³ Amir Bastani⁴

1) School of Advanced Technologies in Medicine. Isfahan University of Medical Sciences, Isfahan, Iran
2) School of Advanced Technologies in Medicine. Medical Image and Signal Processing Research Center, Isfahan University of Medical Sciences
3) Isfahan Branch of Iran Language Institude, Isfahan, Iran
4) Medical School, Kermanshah University of Medical Sciences, Kermanshah, Iran

Publication : The 6th International conference of Sciences and Engineering(6icesconf.com)
Abstract :
Human heart, as a vital organ to continue living, can be featured by bioelectrical characteristics. One of the main modalities for monitoring these features is ECG signal. This signal is often combined with artifacts causing noise and baseline wander which make difficulties in clinical diagnosis and beat detection. One of the most important issues is detrending and denoising the signal to have a better clinical identification and beat detection. In this work, we propose a novel automatic method to detrend and denoise ECG signals and detect beat locations in them. We use symlet4 wavelet decomposition in order to detrend and denoise ECG signal and continue with periodic normalization and curve fitting methods to identify beat locations. We tested our algorithm on 47 ECG signal records acquired from MIT-BIH arrhythmia database and obtained the sensitivity of 99.78% and the accuracy of 99.40%, while its time complexity had a low cost for this accuracy level.
Keywords : ECG Analysis; Beat Detection; Wavelet Decomposition; Signal Denoising; Symlet4