Use of Multi-Scale Principal Component Analysis for motion artifact reduction of PPG signals

Abstract

Arterial blood oxygen saturation (SpO2), a vital measure of amount of oxygen that is dissolved in blood, is estimated using commercial pulse oximeter by recording the Photoplethysmographic (PPG) signals. Ever since the invention of pulse oximetry, reliable and accurate estimation of arterial blood oxygen saturation (SpO2) has been a challenging problem for researchers. Mostly inaccurate estimation of SpO2 in a pulse oximeter arises due to the motion artifacts (MA) created in the detected PPG signals by the voluntary or involuntary movements of a patient. We present an MA reduction method based on Multi Scale Principal Component Analysis (MSPCA) technique. MSPCA combines the ability of PCA to decorrelate the variable with wavelet analysis for MA reduction from recorded PPG data. MSPCA computes PCA of wavelet coefficients at each scale followed by combining the results at relevant scales. Experimental result revealed that MSPCA outperformed the basic wavelet based processing for MA reduction of PPG signals and is best suited for pulse oximetry applications.