Efficacy of DWT denoising in the removal of power line interference and the effect on morphological distortion of underlying atrial fibrillatory waves in AF-ECG

Abstract

The objective of this study is to assess the efficacy of Discrete Wavelet Transform (DWT) in the removal of power line (50Hz) interference (PLI). Eighteen mains noise corrupted ECG signals were denoised using thirty two different DWT mother wavelets in order to assess which are the top performing for power line interference cancellation. For com-parative purposes the signals were also denoised using a traditional notch filtering approach and the results assessed using three performance parameters: Signal to noise ratio (SNR), Mean Square Error (MSE) and Signal Correlation Value (SCV). 12 of the 32 wavelet functions utilized for mains inter-ference denoising outperformed the traditional notch filtering approach, with the top four performing wavelets being Daubechies ‘Db10’, Biorthogonal ‘Bior6.8’, DMeyer ‘Dmey’ and Daubechies ‘Db8’, with Db10 producing SNR, MSE and CCV values of 32.50, 5.13x10^− 5 and 0.9995 respectively. This was considerably better than the notch filtering technique which produced comparable results of 25.67, 1.10x10^− 3 and 0.9952 respectively. The second phase of this study assessed the effect that DWT PLI attenuation had on underlying fibrillatory wave morphology. The results indicate that discrete wavelet processing has a negligible effect on underlying fibrillatory waves and is therefore a viable method for mains noise remov-al in ECG analysis of AF patients.

Keywords