Motion Artifact Identification and Removal From Wearable Reflectance Photoplethysmography Using Piezoelectric Transducer

Abstract

Accurate heart rate estimation from reflectance photoplethysmography (PPG) during intense physical activity is difficult given unpredictable motion artifacts (MAs). Recent research has focused on the removal of MAs in PPG signals by considering acceleration signals as noise references. However, this approach does not always effectively remove MAs, as they can be associated with other noise sources. For instance, finger or hand motion reflects upon the wrist, corrupting the PPG signals, but an accelerometer cannot capture this type of MA. Hence, we propose adding a piezoelectric transducer on the wrist to identify noise originated from finger and hand motion. We first detail the developed wearable device that simultaneously measures PPG, acceleration, and piezoelectric pressure signals from the wrist. Then, the effectiveness of piezoelectric pressure signals for MA removal is verified from PPG signals collected under finger tapping and hand folding/unfolding, and from data of 17 healthy subjects undergoing physical activity. MA removal using both acceleration and piezoelectric pressure signals exhibits a mean absolute error of 1.58 beats per minute (bpm), considerably below the error of 3.75 bpm obtained using only acceleration signals. Therefore, measuring piezoelectric pressure signals on the wrist can improve heart rate estimation by providing more information of MAs corrupting PPG signals.