Effective systolic peak detection algorithm using variational mode decomposition and center of gravity

In this paper, we propose a new automated method for determining systolic peaks in the photoplethysmography(PPG) signal. The proposed systolic peak detection method includes four stages. In the first stage, variational mode decomposition (VMD) technique is used for enhancing the PPG signal and suppression of background noises such as baseline wanders, powerline and high-frequency noises. In the second stage, the candidate envelope is obtained using a center of gravity (CoG) technique. In the third stage, the locations of peaks present in the local waves of the envelope are determined by detecting the positive zerocrossing points of the CoG envelope. Finally, the detected peak locations are used for finding true locations of systolic peaks present in the PPG signal using the peak adjustment procedure. The proposed method is validated and tested by using variety of PPG signals. Evaluation results showed that the method achieves an average Se of 99.10%, Pp of 99.77%, DER of 1.13% and ACC of 98.88% for MIT-BIH Polysomnographic database and an average Se of 99.39 %, Pp of 99.67 %, DER of 0.94 %, and ACC of 99.06% for the real-time PPG signals.