Adaptive Algorithms to Optimise Photoplethysmographic Signals in Heart Rate Estimation

J.Y.A. Foo, W.-Y. Leong, S.J. Wilson, and J. Homer (Australia)


Signal processing, motion artefact, poor perfusion, adaptive filter, accelerometry, and wavelet transform


The use of pulse oximetry has been extensively in the medical fields. Besides being the common arterial blood oxygen saturation (SaO2) measure, pulse oximeters can also be used for other timing applications like, heart rate (HR) estimations. However, motion artefact and poor peripheral perfusion are common phenomenon that contaminates its photoplethysmographic (PPG) signals. This paper introduces an approach to address both its intrinsic weaknesses in timing-related application. In this proposed system, accelerometry is used to detect the presence of mild artefacts and then employing adaptive filtering to minimise its effects on the PPG signals. The wavelet denoising algorithm is engaged to suppress the noise component in the PPG signal due to a poorly perfused periphery. For HR estimate comparison, PPG signals are evaluated by comparing their beat-to-beat estimates with the corresponding R-R intervals from an electrocardiogram (ECG). Statistical analyses on the proposed approach show promising results; tidal breathing (r2 =0.98; p<0.05), mild regulated artefacts (r2 =0.67; p<0.05) and poorly perfused finger (r2 =0.58; p<0.05). A decision matrix then adopts the appropriate method to improve the PPG signal-to-noise ratio dynamically. The results attained here demonstrate that the proposed approach has great potential for use in relevant medical applications.

Important Links:

Go Back