A comparative analysis of ECG denoising methods

A common problem in ECG analysis is the removal of unwanted noise that corrupt the signal while recording and can hinder the correct interpretation. To provide high-quality ECG records, many approaches have been developed. This paper introduces a comprehensive survey of filtering methods to cope with the noise artifacts in the ECG signals in order to find the best-suited filter for each type of noise. For this purpose, we used 47 half hour signals coming from adding artifacts to clean ECG records taken from real clinical databases. We calculated the Signal to Noise Ratio (SNR) and the Root Mean Square Error (RMSE) after filtering and then a statistical test was used to select the most efficient method. The noises that were filtered are Power Line Interference (PLI), Baseline Wander (BW), Electrod Motion Artifacts (EM) and Muscle Artifacts (MA). Empirical Mode Decomposition was the best method to filter PLI noise. BW, MA and EM were removed to a large extent using Wavelet+Adaptive filtering method.