Denoising of ECG Signals Using Wavelet Transform and Principal Component Analysis

Abstract

Electrocardiogram (ECG) is used for the analysis of the electrical activity of the heart. Processing of these ECG signals is thus important for detecting the presence of any abnormalities or onset of diseases. Distortion due to noise results in changes in the amplitude and frequency of the signal obtained, resulting in an inaccurate diagnosis. Various denoising techniques have been employed to remove noise or extract critical information from the signal. This paper attempts to study Discrete Wavelet Transform (DWT), Principal Component Analysis and associated parameters such as wavelet family, thresholding method and component analysis rule. Through results analysis, it was determined that Symlet wavelet family provided the best results for noise removal of the signals with its high SNR and PSNR values and it was observed that the scaling function of a symlet resembles that of a clean ECG signal.

Keywords: Electrocardiogram, Denoising, Discrete Wavelet Transform, Principal Component Analysis, Symlet