Combining accelerometry-based motion artifact cancellation with pulse oximetry processing

Abstract

This paper proposes a combination of motion artifact (MA) cancellation with pulse oximetry processing. The objectives were to develop motion-resistant pulse oximetry processing relying on using accelerometry-based MA cancellation, and to explore the performance of the proposed processing method in providing arterial oxygen saturation () when the red and infrared (IR) photoplethysmogram (PPG) were corrupted by a MA. A PPG acquisition system was designed and constructed in the laboratory for the experiments. A three-axis accelerometer was utilized to detect the motion of a finger probe. An acceleration signal corresponding to the longitudinal direction of the finger was used to generate a motion reference signal for active noise cancellation using an adaptive filter. A Savitzky-Golay filter was applied to obtain the normalized first derivative of red and IR PPGs, and to generate a motion reference signal. The MA was eliminated from the normalized first derivative of red and IR PPGs using an adaptive filter, of which the adaptive algorithm was a recursive least square (RLS) method. The appropriate filter order was 16. After the MA cancellation, the normalized first derivative of the red and IR PPGs were expected to be free of MA, and they represented the arterial light absorption. The ratio of red to IR absorption () was computed and converted to the  using the saturation equation obtained from the calibration. The experimental results showed that the theoretical concept for combining the accelerometry-based MA cancellation with the pulse oximetry processing was practical. The proposed processing method helped reduce the error of the  measurement when a MA was present, and it had more resistance to MA than conventional pulse oximetry processing.