A Wavelet Denoising and Teager Energy Operator-Based Method for Automatic QRS Complex Detection in ECG Signal

摘要

The electrocardiogram is an important tool that is widely used for diagnosis of many cardiovascular diseases. In this context, QRS complex detection is a very crucial step in the ECG diagnosis system. The major aim of this work is to develop a novel method for QRS complex detection under various ECG signal morphologies as well as under different ECG recording conditions, including numerous noise sources and varying QRS waveforms. The proposed algorithm is based principally on the stationary wavelet transform (SWT) and Teager energy operator (TEO). In our scheme, SWT is first used for ECG signal preprocessing and QRS complex frequency content localization. Subsequently, a novel process for R peak detection based on TEO and a moving average (MA) filter is introduced. More precisely, SWT is coupled with TEO and the MA filter to construct a smoothed detection mask. Then, after the mask segmentation and adaptive thresholding steps, R peak times are identified using the maxima detected on the created mask and employing a reference ECG signal. At this stage, efficient decision rules are applied for reducing the number of false alarms. In the experiments, we validate the proposed method on the well-known annotated MIT-BIH arrhythmia database (MITDB). The experimental results show that the newly proposed algorithm provides satisfactory detection performances compared to the recent state-of-the-art methods, with an average sensitivity of 99.84%, average positive predictivity (P+) of 99.87%, detection error rate of 0.30% and an overall detection accuracy of 99.70%. Also, the proposed method presents a low computational time complexity with an average processing time of 12 s on each ECG record from MITDB.