Effect of Different Signal Processing Techniques on a Calibration Free Pulse Oximeter

A pulse oximeter is a medical device that measures pulse rate and oxygen saturation levels of a person non-invasively. Monitoring of health parameters of a patient requires a device that gives accurate results in all conditions and is not affected by artifacts arising from environmental and physiological factors. One of the noise factors we focus in this paper is motion artifacts. These artifacts affect the quality of signal acquired by the sensor leading to imperfect results. In this paper, four digital filtering techniques, which include Exponential Moving Averaging, Savitzky-Golay, Independent Component Analysis and Discrete Kalman filter, are implemented to remove artifacts from a noisy photoplethysmographic signal acquired by shining lights in the range of Red and IR wavelengths on the fingertip of a person. Pulse rate is determined in the time domain by detecting peaks of the signal while blood oxygen levels are measured using Beer-Lambert Law via a calibration-free method on the artifact-free PPG signal. The results obtained are then compared with a commercial pulse oximeter available in the market and performance of different filtering techniques on a PPG signal is evaluated.