Performance of Wavelet Energy Gradient method for QRS detection

In this paper, we illustrate the performance of Wavelet Energy Gradient method for QRS detection of ECG signal which is corrupted with five different types of noise and artefacts. In order to validate the robustness of this method, important noise sources such as powerline interference, baseline shift due to respiration, abrupt baseline shift, electromyographic (EMG) interference were combined with ECG signal and it was found that this method exhibits reliable detection with an average detection error rate as low as 4% and average sensitivity of 99.5%. The performance is also evaluated by adding White Gaussian Noise (WGN) at various noise (SNR) levels; 60dB to 5dB and this method is found to be tolerant down to 35 dB. The experimentation is carried out for four types of records; Normal, Premature Ventricular Contraction (PVC), Right bundle branch block beat (RBBB) and Left bundle branch block beat (LBBB) from the standard MIT-BIH arrhythmia database.