Automatic SWT Based QRS Detection Using Weighted Subbands and Shannon Energy Peak Amplification for ECG Signal Analysis Devices

In this paper, we present a straightforward automatic QRS complex detection method for electrocardiogram (ECG) signal analysis applications. The proposed method consists of stationary wavelet transform (SWT) for suppressing low- and high-frequency noises and extracting QRS complexes, amplitude thresholding to suppress the effect residual noise components, Shannon energy based peak amplitude normalization, negative zero-crossing for detecting peaks candidate smoothed QRS complex waveform and peak correction for determining true R peaks in the ECG signal. On the standard MIT-BIH database, our method had an accuracy of 99.50%, sensitivity of 99.69%, and a positive predictivity of 99.81 %. The proposed method outperforms other existing methods which included sets of amplitude-and duration-dependent thresholds to include or reject missed R peaks and noise peaks, respectively that may not work in practise for the case of QRS complex with irregular rates and long-pause between two consecutive QRS complexes.