A Computational System to Optimise Noise Rejection in Photoplethysmography Signals During Motion or Poor Perfusion States

Photoplethysmography (PPG) signals can be used in clinical assessment such as heart rate (HR) estimations and extraction of arterial flow waveforms. Motion artefact and/or poor peripheral perfusion can contaminate the PPG during monitoring. A computational system is presented here to minimise these two intrinsic weaknesses of the PPG signals. Specifically, accelerometers are used to detect the presence of motion artefacts and an adaptive filter is employed to minimise induced errors. Zero-phase digital filtering is engaged to reduce inaccuracy on the PPG signals when measured from a poorly perfused periphery. In this system, a decision matrix adopts the appropriate technique to improve the PPG signal-to-noise ratio dynamically. Statistical analyses show promising results (maximum error < 7.63%) when computed HR is compared to corresponding estimates from the electrocardiogram. Hence, the results here suggest that this dual-mode approach has potential for use in relevant clinical measurements.