R Peak Detection Based on Wavelet Transform and Nonlinear Energy Operator

Abstract

The Electrocardiography is a graphical representation of the heart’s electrical activity. In this paper, we present an efficient algorithm for QRS complex detection in the Electrocardiogram (ECG) signal. The proposed algorithm is based on a combination of the Shift Invariant Wavelet Transform (ShIWT), a Nonlinear transform called Nonlinear Energy Operator (NEO) and a simple thresholding function followed by some decision rules for accurate R peak detection. In our scheme, ShIWT was used to filter out the raw ECG signal and the NEO was applied to highlight the QRS complex patterns. Finally, after simple thresholding stage, R peak time positions on the filtered ECG signal can be detected accurately with the help of efficient decision rules. The experimental results and tests carried over real ECG signals taken from the MIT-BIH Arrhythmia Database (MITDB) show that our proposed approach gives a comparable or higher detection performances against the state of the art techniques with an average Sensitivity (Se) of $99.76\%$, average Positive Predictivity (P+) of $99.77\%$ and a Detection Error Rate (DER) of $0.47\%$.