Abstract. This paper proposes an alternative method for pulse oximetry processing based on time-frequency analysis. Instead of using the fast Fourier transform (FFT), the continuous wavelet transform (CWT) together with the median calculation was proposed for finding the frequency components of the normalized first derivative red and infrared photoplethysmographic signals (PPG). In order to estimate the arterial oxygen saturation (% 2 ), the ratio of each red CWT modulus to each corresponding infrared CWT modulus was computed at each frequency, and then converted to the oxygen saturation by using the saturation equation obtained from calibration. All of the oxygen saturation data were plotted on a histogram. The greatest oxygen saturation with the most occurrences was considered to represent the % 2 . The experiments were held to evaluate the performance of the proposed processing method compared with the conventional pulse oximetry (CPO) processing. The red and infrared PPGs were acquired from left and right index fingers simultaneously by using the PPG measurement system constructed in the laboratory. During measurement, the left index finger was stationary, while the right index finger was performed the finger bend to induce the motion artifact (MA) in transient and periodic manners. All detected PPGs were processed by the proposed method and the CPO processing. The % 2 obtained from different processing methods and positions were compared. Experimental results showed that the proposed method was more resistant to the MA than the conventional method.

No abstract