Sensitivity analysis of near infrared glucose absorption signals: toward noninvasive blood glucose sensing

Noninvasive blood glucose monitoring is a long pursued goal in clinical diagnostic. Among several other optical methods, near infrared absorption spectroscopy is the most promising one for the noninvasive application to date. However, realization has not been achieved. A major obstacle is the low signal-to-noise ratio pertinent to physiological blood glucose measurement using the near infrared absorption technique. Sensitivity analysis of aqueous glucose absorption signals was performed in the combination band region and in the first-overtone region. The analysis involved quantification of both glucose absorption signal and the corresponding spectral noise within a particular wavelength region. Glucose absorption band at 4430cm(-1) (2257nm) in the combination band region was found to give an order of magnitude higher signal-to-noise ratio than the strongest band in the first-overtone region. A Fourier-filtering algorithm was applied to the raw absorbance data to remove some of the unwanted spectral variations. With simple peak-to-peak analysis to the Fourier-filtered absorbance data, repeatability of less than +/-0.5mmol/L was achieved. In addition, effects of temperature variations on the absorption spectra were studied. The effects of sample temperature were compensated with the application of the Fourier filter.