An Approach For Measurement Of Spo2 In Presence Of Motion Artifact

Abstract

Obtaing accurate percentage of oxygen saturation (SpO2) using finger-probe based newlinepulse oximeter (PO) sensor is reliant on both Red and Infrared Photoplethysmogram (PPG) newlinesignal. In realistic environment conditions, PPG signals are stained by a motion artifact newlinesignals which are produced due to body disturbance or finger motion/fluctuation. To rectify newlinethis motion artifact interference, the reason for the corruption of Photoplethysmogram signals newlineby the motion artifact signals is scanned. The motion artifact signals are established to newlineperform like an additive noise which gives false reading. Motion and noise artifacts (MNAs) newlineenforce bounds that are threshold on the applicability of the Photoplethysmogram, newlinespecifically in the sleep disorder and ambulatory monitoring (AM) detection. Motion and newlinenoise artifacts can garble Photoplethysmogram, triggering inaccurate valuation of newlinephysiological outcomes such as a saturation of oxygen (SpO2) and heart rate (HR). newlineTherefore, it is challenging to select the right kind of sensor device for measuring these newlinesignals. For overcoming research challenges, this work conducted extensive survey, newlineidentified challenges and finally selected MAX 30100 sensor as a final choice for newlinemeasurement. Then, this work aimed at generating SpO2 raw data using MAX 30100 sensor newlinefor both with and without motion artifacts scenarios. For generating clean SpO2 data more newlinecare is taken. Then, using known motion pattern we have taken data for obtaining raw data newlinewith motion artifacts for different individuals. Further, this thesis presented a novel hybrid newlineapproach for detection of noise and motion artifacts. Firstly, this work presents an accurate newlineSpO2 measurement model. Secondly, present an adaptive filter and adaptive threshold model newlineto detect artifact and obtain derivative of correlation coefficient for labeling artifacts newlinerespectively. Lastly, Enhanced Support Vector Machine (ESVM) Model is presented to newlineperform classification. Experiment is conducted on both real-time and simulated dataset. Our newlinehybrid approach attains significant performance in term of accuracy, sensitivity, specificity newlineand positive prediction. Further, experiment is conducted to evaluate sleep apnea detection. newlineThe outcome shows the proposed hybrid model attain significant performance in term of newlineaccuracy, sensitivity, specificity and positive prediction. newline