A novel method for pulse onset detection on pliotoplethysmogram using changes of waveform trend

Abstract

Photoplethysmogram (PPG) is an important physiological signal for evaluating the status of arterial wall, such as calculating markers like pulse wave velocity and pulse arrival time to reflect arterial stiffness. Detecting the onset points is an important prerequisite for analyses of PPG. In this paper, we proposed a novel algorithm using trend change of PPG waveform to detect its pulse onset points. This idea raised from the physiological phenomenon that due to the arrival of arterial pulse, the trend of PPG waveform changes sharply on both sides of PPG onset points from a decreasing trend in the previous end-diastolic phase to a sharp rising trend in the coming early systolic phase. In our algorithm, a published method was adopted to detect the peaks of 1 st derivative wave of PPG. And then, in the neighborhood of each 1 st derivative PPG peak point, the waveform trend changes between two sides of each point in this neighborhood was calculated, and the onset point was defined as the one with biggest positive trend change. The trend on one side of a target point was calculated from the arc tangent angles between the target point and its nearby points on this side. To evaluate the performance of the proposed algorithm, data collected on 15 healthy elderly subjects in a previous exercise study was analyzed. The results showed that the proposed algorithm achieved at a sensitivity of 99.69%, a positive predictivity of 99.24% and a failed detection rate of 1.07%. Our results demonstrated that the proposed method was robust to noises during exercise and was able to detect PPG onset points of these subjects at a satisfactory accuracy. Since the algorithm is also low in computing cost, it is promising to be used real time in ambulatory/wearable systems in the future.