Comparison of foot finding methods for deriving instantaneous pulse rates from photoplethysmographic signals

Abstract

The suitability of different methods of finding the foot point of a pulse as measured using earlobe photoplethysmography during stationary conditions was investigated. Instantaneous pulse period (PP) values from PPG signals recorded from the ear in healthy volunteer subjects were compared with simultaneous ECG-derived cardiac periods (RR interval). Six methods of deriving pulse period were used, each based on a different method of finding specific landmark points on the PPG waveform. These methods included maximum and minimum value, maximum first and second derivative, ‘intersecting tangents’ and ‘diastole patching’ methods. Selected time domain HRV variables were also calculated from the PPG signals obtained using multiple methods and compared with ECG-derived HRV variables. The correlation between PPG and ECG was greatest for the intersecting tangents method compared to the other methods (RMSE = 5.69 ms, r2 = 0.997). No significant differences between PP and RR were seen for all PPG methods, however the PRV variables derived using all methods showed significant differences to HRV, attributable to the sensitivity of PRV parameters to pulse transients and artifacts. The results suggest that the intersecting tangents method shows the most promise for extracting accurate pulse rate variability data from PPG datasets. This work has applications in other areas where pulse arrival time is a key measurement including pulse wave velocity assessment.