Key points are not available for this paper at this time.
Non-invasive glucose monitoring represents a significant advancement in diabetes management and treatment as non-painful alternatives than finger-sticks tests. After developing an integrated Raman spectral system with a 785 nm laser, this study systematically explores the application of in vivo Raman spectroscopy for quantitative, noninvasive glucose monitoring. In addition to observing characteristic glucose spectral information from a mouse model, a strong spectral correlation was also recognized with the blood glucose concentration . The glucose fingerprint information detected from the nailfolds of 30 human volunteers exhibited concentration dependent changes, especially when the intraspectrum intensity ratio was calculated between 1125 cm −1 and 1445 cm −1 to monitor normalized differences in the glucose Raman band. Furthermore, by accounting for all intersubject variations observed in the acquired spectral features, a particle swarm optimization-backpropagation artificial neural network (PSO-BP-ANN) model was proposed for linking measured Raman information with actual glucose concentrations quantitatively. Following model training and testing, the prediction accuracy of the PSO-BP-ANN model was evaluated using 12 spectra acquired from an additional three volunteers. Statistical evaluations indicated that the proposed methodology may have a good application potential for in vivo transcutaneous spectral glucose monitoring.
Liu et al. (Sun,) studied this question.