Motion artifact removal method for PPG signals based on CEEMDAN-LMS joint filtering

Photoplethysmography (PPG) signals, measured through pulse oximeters embedded in wristband devices, commonly utilized for heart rate monitoring. These wearable sensors can assist in the early detection of abnormalities by providing healthcare professionals with a basis for taking preventive measures. However, in practical scenarios, motion artifacts significantly interfere with PPG signals, posing a major challenge to accurate monitoring in dynamic environments. To improve the quality of PPG signals and enhance their reliability in health monitoring, this study proposes a motion noise elimination method based on the combination of Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Least Mean Squares (LMS) filtering. The performance of this method is evaluated by comparing the signal-to-noise ratio (SNR) of reconstructed PPG signals processed with traditional EMD and CEEMD algorithms. The findings indicate that the CEEMDAN-LMS method is better suited for handling such signals, demonstrating significant advantages in improving PPG data quality. Additionally, the study explores Multi-Scale Permutation Entropy (MPE) in CEEMDAN decomposition. By calculating the MPE values of Intrinsic Mode Functions (IMFs), useful signals and noise components can be effectively distinguished, further optimizing the signal separation process.