Complex technique for wavelet filtering of pulse wave signal

Abstract

The article is devoted to the research of a comprehensive technique for digital filtering of the pulse wave signal in the presence of various physiological artifacts, such as baseline wander and motion artifacts. The proposed method of wavelet filtering of a pulse wave signal from physiological artifacts based on discrete decomposition into orthogonal wavelets includes sequential procedures for digital processing: multiscale wavelet transform; modification of detail coefficients of wavelet decomposition based on thresholding; reconstruction of the pulse wave signal based on the original approximation coefficients and modified detail coeffi cients using the inverse wavelet transform. A comparative analysis of the proposed methodology with existing approaches to filtering pulse waves, such as moving average filtering, median filtering, bandpass frequency filtering, was carried out. To obtain quantitative characteristics for evaluating the filtering efficiency, we used simulation of a pulse wave with the presence of interference and noise of various intensity and nature of occurrence. The studies carried out in this work have shown that multiscale wavelet transformations of the pulse wave signal provide the least distortions when filtering motion artifacts in comparison with classical approaches based on temporal or spectral transformations, while the advantages of multiscale wavelet analysis are most noticeable in conditions of increased noise.