On the use of Wavelet Transform based Adaptive Filtering for de-noising of Pulse Oximeter signals

ABSTRACT

Monitoring patient's blood oxygen saturation (SpO(2)) levels using pulse oximeter is important to physician. SpO(2) is also one of the major parameter that is being monitored to assess the respiratory health in Covid-19 infected patients during the ongoing pandemic. In pulse oximeters, the motion artifacts (MA), due to voluntary or involuntary movement of patient, will disturb the morphology of the photoplethysmographic (PPG) signals acquired through a finger/forehead sensor resulting in inaccurate SpO(2) values. The current work is focused on an efficient adaptive filtering method for MA reduction, which uses a wavelet reconstructed secondary MA noise as reference signal. It eliminates the use of an external sensor to be employed for estimating MA signal. This method while reducing the MA restored the PPG morphology and respiratory components facilitating accurate estimation SpO(2), heart rate (HR).