Adaptive low-power wrist SpO2 monitoring system design using a multi-filtering scheme

Abstract

Objective:As most existing medical pulse oximeters conduct the monitoring of blood oxygen saturation (SpO2) at fingertips, the compliance of subjects cannot be ensured. In this paper, we aim to design an SpO2 monitoring system on the wrist, where the thick cortex and unclear vessels make the original collected signals affected by complex noises. Methods:We render an adaptive low-power wrist SpO2 monitoring system, where a multi-filtering scheme based on thresholding, Gaussian filtering and Butterworth filtering is proposed to process the original collected signals. Besides the wrist-worn monitoring terminal, a database is also established on the server for data display and management. The resulting SpO2 values were compared to the professional medical pulse oximeters Yuwell YX306 and Philips DB18. Results:With the multi-filtering scheme, the measurement results of the proposed system matched the medical pulse oximeters well, with average relative errors of 0.6175% compared to the YX306 and 0.555% compared to the DB18 in 400 SpO2 measurements, which were far lower than the tolerance of commercial devices under the ISO80601-2-61 standard. The power consumption of the proposed system was only 2 mW on average. Conclusion:The multi-filtering scheme can remove most noises in the original collected signals, and the proposed system gets superior performances on SpO2 measurement accuracy and power consumption. Significance:The proposed system, designed using the multi-filtering scheme, is able to conduct a stable and effective wrist SpO2 monitoring, with the subject’s compliance ensured. It achieves low power consumption and can be used for long-time continuous SpO2 monitoring.