Ensemble empirical mode decomposition applied for PPG motion artifact

Abstract

This study evaluates the performance of the ensemble empirical mode decomposition (EEMD) filtering applied to the vertical movement motion artifact (MA). The evaluation of the filtering algorithm is investigated by the heart rate (HR) frequency assessment calculated by the fast Fourier transform (FFT), the intrinsic mode function (IMF) selection algorithm, and the periodogram. In this study, the results show that the raw photoplethysmograph (PPG) signal has some defects where the MA is applied, about 0.6 Hz of the vertical movement. Another result by utilizing the EEMD filter, it can be seen that, even though the period when the MA is activated, the HR frequency is relatively stable, about 1.3 Hz, by evaluating the time-frequency and maximum dominant frequency for a small period of the windowed signal.