Abstract
In this study, to eliminate the motion artifacts contaminating photoplethysmograpic (PPG) signals, a clip-type earlobe photoplethysmography sensor was improved and a noise-reduction algorithm based on a multiple-regression model was examined. Motion artifacts are thought to be caused by a change in the flow of blood. We thought that such motion artifacts could be reduced considering the acceleration change during body motion. In this study, an acceleration sensor was integrated with a commercially available clip-type earlobe PPG sensor. For the first step, typical motion artifact waveforms were calculated by choosing every motion artifact waveform, arranging the peak time and averaging. In the second step, to make a multiple regression analysis between averaged acceleration and averaged motion artifact waveform, regression coefficients were calculated. Finally, an original PPG signal with motion artifact was filtered using an individually adjusted multiple regression model. In experiments, five subjects were asked to walk and run on a treadmill to verify this filtering method. Results show that our new motion artifacts reduction algorithm is proven to be effective for decreasing the influence of body motion, and the detection accuracy of the number of pulsations has improved up to 13.3%.
