The Study On Theories And Experiments Of Temperature Amendment Method In The Use Of Near-Infrared Spectroscopy

Noninvasive measurement of human blood glucose concentration with near-infrared spectroscopy has becoming more and more focused in the area of biomedical engineering. Nevertheless, one of the critical obstacles that lie in the clinical realization of noninvasive human blood glucose concentration is the uncontrollable and unpredictable temperature fluctuation of human body. As the temperature varies from time to time, the measurement spectroscopy will also be changed, and thus adversely affect the measurement precision. Therefore, this paper focuses on the aspect of temperature amendment of near-infrared spectroscopy.In this dissertation, the characteristic of temperature influence of near-infrared spectroscopy is discussed. Based on Lambert-Beer's Law and the temperature feature of molar absorption coefficientε, the correlation between the changes of near-infrared spectroscopy and the changes of temperature and glucose concentration is established. As the sample's temperature and glucose concentration vary, the sample'sΔA signal can be regarded as consisting two parts: one solely related to temperature shifts:ΔA T (λ), the other one related only to glucose concentration changes:ΔA C (λ). Results of the forward temperature effect experiment show that: in the wavelength area ranging from 1200nm till 1700nm, 1℃of temperature changes can induce great effect on the spectroscopy, which usually equals to the effect of dozens or several hundreds of mmol/L glucose concentration changes. This will severely affect the precision and feasibility of noninvasive glucose concentration measurement.Based on the characteristics of glucose reference-wavelength and under the direction of Yun CHEN, a spectroscopy temperature melioration method is established. In the glucose reference-wavelength point, the light intensity is insensitive to the changes of glucose concentration and thus at this wavelength the changes of light intensity can be totally regarded as the consequence of temperature noises. Therefore, the measurement temperature of samples can be calculated using the changes of light intensity of reference-wavelength point. Results of the temperature calculation experiment show that, the average reference-wavelength calculation error of this method is less than 0.04℃. In regard of this, signalΔA T (λ) which caused by temperature shifts can be measured in all wavelength band, and can therefore obtain the glucose specific signalΔA C (λ). Processing the spectroscopy of aqueous glucose solution of different temperatures using this method, the noise signal which resulted from temperature shifts can be eliminated. After this spectroscopy modification, the spectral lines can depart with each other disciplinarily according to different glucose concentration.Finally, the effect of this reference-wavelength spectroscopy temperature amendment is evaluated. The calibration model is established using aqueous glucose solution under 30℃. The prediction sample spectroscopy is then measured under different temperature (32℃till 40℃). Then the above reference-wavelength spectroscopy temperature amendment method is introduced to modify prediction sample's original spectroscopy. Then, based on the same 30℃calibration model, the prediction sample spectroscopy before temperature correction and after temperature correction are predicted respectively. The result shows that, after temperature amendment, the glucose concentration prediction precision is enhanced by nearly 250 times.