Continuous signal quality estimation for robust heart rate extraction from photoplethysmographic signals

This study presents a novel method for estimating the signal quality of photoplethysmographic (PPG) signals. For this purpose a robust classifier is implemented and evaluated by using finger- and inear-PPG. A new procedure is proposed, which uses feature reduction to determine the Mahalanobis distance of the PPG-pulses to a statistical reference model and thus facilitates a robust heart rate extraction. The evaluation of the algorithm is based on a classical binary classification using a manually annotated gold standard. For the finger-PPG a sensitivity of 86 ± 15 % and a specificity of 94 ± 13 % was achieved. Additionally, a novel classification method which is based on a continuous signal quality index is used. Pulse rate estimation errors greater than 5 BPM can be detected with a sensitivity of 91 ± 13 % and a specificity of 91 ± 15 %. Also, a functional correlation between the signal quality index and the standard deviation of the pulse rate error is shown. The proposed classifier can be used for improving the heart rate extration in pulse rate variability analysis or in the area of mobile monitoring for battery saving.