OPRA: A Fast On-Line Signal Quality Estimator for Pulsatile Signals

Automatic evaluation of physiological signals is considered an important research topic which helps to significantly facilitate the work of physicians and involved medical staff. Moreover, reliable on-line detection of various pathophysiologies is also of great interest especially in intensive care units or home monitoring settings. Special care has to be taken when processing pulsatile signals like the photoplethysmogram (PPG) which are particularly affected by motion artifacts. This paper presents a novel algorithm based on a stochastic gradient ascent procedure which allows for a reliable on-line PPG signal quality estimation that can be implemented on mobile sensor devices with limited processing power. The performance of the algorithm is analyzed on data drawn from a conducted field study comprising 20 subjects exercising different experiments as well as 10 records downloaded from Physionet database. The proposed method achieves reasonable results without the need for complex parameter adjustments and performs well in changing measurement environments.