Determination of NIR informative wavebands for transmission non-invasive blood glucose measurement using a Fourier transform spectrometer

Abstract

Non-invasive blood glucose measurement using near infrared (NIR) spectroscopy relies on wavebands that provide reliable information about spectral absorption. In this study, we investigated wavebands which are informative for blood glucose in the NIR shortwave band (900∼1450 nm) and the first overtone band (1450∼1700 nm) through a specially designed NIR Fourier transform spectrometer (FTS), which featured a test fixture (where a sample or subject’s finger could be placed) and all-reflective optics, except for a Michelson structure. Different concentrations of glucose solution and seven volunteers who had undergone oral glucose tolerance tests (OGTT) were studied to acquire transmission spectra in the shortwave band and the first overtone band. Characteristic peaks of glucose absorption were identified from the spectra of glucose aqueous solution by second-order derivative processing. The wavebands linked to blood glucose were successfully estimated through spectra of the middle fingertip of OGTT participants by a simple linear regression and correlation coefficient. The light intensity difference showed that glucose absorption in the first overtone band was much more prominent than it was in the shortwave band. The results of the SLR model established from seven OGTTs in total on seven participants enabled a positive estimation of the glucose-linked wavelength. It is suggested that wavebands with prominent characteristic peaks, a high correlation coefficient between blood glucose and light intensity difference and a relatively low standard deviation of predicted values will be the most informative wavebands for transmission non-invasive blood glucose measurement methods. This work provides a guidance for waveband selection for the development of non-invasive NIR blood glucose measurement.

Near-Infrared Spectroscopy in Bio-Applications

“…Therefore, further advances are needed to develop a more robust and universal prediction model. Further studies from other groups aimed at determining the informative wavelengths in the NIR region for non-invasive blood glucose prediction should be noted, e.g., Yang et al [24], or Suryakala and Prince who used PLS regression and principal component regression (PCR) models for this purpose [25]. Non-linear regression by means of artificial neural network (ANN) has also been evaluated and compared with PLS regression by Jintao et al [26].…”

Ultrasound-modulated optical glucose sensing using a 1645 nm laser

“…To achieve strong glucose absorption contrast, we used a 1645 nm laser based on the measured absorbance spectra. This wavelength lies in the first overtone band (1,500-1,800 nm) known to have primary absorption peaks of glucose in the NIR region 56,57 . Two different ultrasound transmission modes, bursts with a 10% duty cycle and continuous waves, were used to modulate diffusive photons in a 20-mm-thick phantom, and the resultant modulation depths were compared.…”

Noninvasive Electromagnetic Wave Sensing of Glucose

“…Besides this, different combinations of pretreatment methods, including first derivative, second derivative, and vector normalization, were also investigated. In their study, PLS achieved a better performance than ANN did with a lower RMSE and a higher R. Yang and co-workers investigated the informative bands in shortwave and first overtone regions [ 57 ] by a homemade NIR Fourier transform spectrometer. All-reflective optics with an Offner relay lens were employed to reduce axial chromatic and magnification chromatic aberrations.…”