A Brief Comparison of Adaptive Noise Cancellation, Wavelet and Cycle-by-Cycle Fourier Series Analysis for Reduction of Motional Artifacts from PPG Signals

Abstract

The accuracy of Photoplethysmographic signals is often not adequate due to motional artifacts induced in the recording site. Over recent decades there has been a widespread effort to reduce these artifacts and different methods are used for this aim. Nevertheless there are still some contradictory results reported by different methods about their effectiveness in artifact reduction. In this paper, we aim to compare three of established methods for PPG noise reduction on a unique dataset. Among different reported methods, we have chosen Adaptive Noise Cancellation (ANC), Discrete Wavelet Transform (DWT) and a newly developed method Cycle-by-cycle Fourier Series Analysis (CFSA) for denoising. To evaluate the effectiveness of mentioned methods, the Heart Rate (HR) estimated from denoised signals by each method has been calculated and compared to the extracted parameters from reference PPG signal which is artifact free. Results indicate that Cycle-by-cycle Fourier Series Analysis method (CFSA) gives the closest results to the results obtained from Reference signal and the Adaptive Noise Cancellation (ANC) method results in more accurate estimation of HR than DWT Denoising method.