An Improved Method Based on Wavelet Threshold to Eliminate the Noise of Blood Oxygen Detection Signal

Capillary refill time (CRT) is the time needed to restore the original color of the distal capillary bed after compression. The time can reflect the microcirculatory state of the human body. However, it is easy to produce noise due to the limitation of the measurement mode. Blood oxygen detection signal is the data obtained by capillary filling time measuring instrument, and it is reflected by the form of signal wave. Due to special measuring means, the signal wave has the characteristics of three states. However, because of this characteristic, the existing de-noising algorithms cannot meet the needs of de-noising blood oxygen detection signals. It is found that the signal noise ratio(SNR) of reconstructed signals is low, which cannot well retain useful information. It is not conducive to the subsequent analysis of blood oxygen detection signals. In order to fit the three states characteristics and prevent signal distortion while maintaining high SNR, this paper presents a method to eliminate the noise of blood oxygen detection signal. Firstly, the measured blood oxygen detection data is plotted as signal wave image, and the signal is preliminarily processed by the improved segmentation wavelet threshold method, and then further denoised according to the adaptive advantages of Hilbert Huang Transform(HHT). Through the results of clinical data experiments, this method can effectively remove the noise in blood oxygen detection signal samples.