Measurement study of oxygen saturation based on D-value estimation of dynamic spectrum

Dynamic spectroscopy (DS) is based on the principle of the photoelectric pulse wave to detect the blood component concentration, which can eliminate the influence of individual difference and measurement conditions in theory by using the absorbance influenced of artery distention degree. Non-invasive measurement of human oxygen saturation is achieved by dynamic spectral transmission method. During the experiment, in-vivo measurements are carried on 62 patients in the intensive care units(ICU) and their spectroscopic data are collected by the high sensitivity type fiber optic spectrometer, then the spectral data collected by spectrometers should be saved in the computer. D-value estimation method is used to extract the dynamic spectrum, the wavelength range of which is 606.44∼991.11nm. The value of arterial blood gas analysis is regarded as the reference to establish the partial least square calibration model between the oxygen saturation and the dynamic spectrum data, 41 groups are selected as calibration set, while 21 groups are selected as prediction set, the correlation coefficient between prediction set and blood gas analysis values is 0.7711 and the relative error is ±0.0198, while the result of the monitor is 0.6770 and ±0.0379, which shows that the correlation coefficient between D-value estimate results and true value is partially improved and the measurement errors of the model compared with the traditional pulse blood oxygen monitor have significantly reduced, the data becomes more stable. Measurement results show that using the high sensitivity type fiber optic spectrometer to collect spectroscopic data and D-value estimate is used to extract dynamic spectrum can eliminate different errors, the spectral data can be made full use of and the signal to noise ratio SNR) can be improved, thus effectively reduce the measurement errors of the oxygen saturation, which provides a new path for NIR spectroscopy non-invasive measurement of blood components.