Assessment of photoplethysmogram signal quality using morphology integrated with temporal information approach

Photoplethysmograph (PPG) signal measured from wearable devices for tele-home healthcare is often corrupted by motion artifacts which often cause false extraction of physiological features and lead to erroneous medical decision for monitoring. In this paper we propose an innovative method which combines the morphological characteristics with temporal variability information in the signal series to assess the signal quality and to reject the meaningless segments that are significantly contaminated by artifacts aiming at improving the accuracy of derived vital physiological features. Experimental results using PPG signals collected in our lab demonstrate that our mechanism can achieve an accuracy of 98.92%.