Mid-infrared Spectroscopic Detection Of Glucose

Glucose(C₆H₁₂O₆)is an important organic molecule that is closely related to human health.Diabetes is a metabolic disease caused by defects in insulin secretion or impaired biological action of insulin,which is characterized by the fact that the blood glucose concentration of the human body exceeds the normal range of human body,resulting in chronic damage to relevant tissues of the human body,especially on the eyes,kidneys,heart,blood vessels,and nerves.Currently,there is no effective curing method for diabetes,and the existing common blood glucose monitoring method is based on chemical measurement on blood droplets.Therefore,development of non-invasive method for continuous blood glucose measurement is of great significance.Due to the presence of strong fingerprint absorptions of glucose,mid-infrared spectroscopy has become a promising way for nondestructive detection of blood glucose.Prior mid-infrared spectroscopic studies on glucose detection often encounter the problems of insufficient fingerprint spectral features and background interference.In this thesis,we conduct an in-depth investigation on characteristic fingerprint absorptions of glucose in the mid-infrared region.Detection of glucose molecules is demonstrated in both aqueous solution and serum solution by using two fingerprint absorption features at 8-10μm and 3.4-3.7μm,respectively.Our research work and results are as follows:（1）Selective fingerprint absorptions of glucose are characterized in water solutions,and detection of glucose concentration in the range of 0～400 mg/dl and 0～1000 mg/dl is demonstrated.Glucose samples were prepared by using deionized water as the dissolvent.Our infrared spectral measurements showed that,besides the commonly reported absorption features at 8-10μm wavelengths,there is an extra fingerprint absorption at 3.4-3.7μm,which offers a new spectral feature for glucose detection.Our data analysis based on a partial least squares regression（PLSR）model showed that the coefficients of determination（R~2）are larger than 0.99 for the two spectra features in the ranges of 0～1000 mg/dl and 0～400 mg/dl.The root mean square errors（RMSECV）are11.85-15.05 mg/dl and 3.48-3.92 mg/dl for the measurement ranges of 0～1000 mg/dl and 0～400 mg/dl,respectively.The glucose detection performance using 3.4～3.7μm absorption feature is slightly better than that of using 8～10μm absorption feature.（2）On the basis of the two fingerprint absorption features at 8-10μm and 3.4-3.7μm as found in water glucose solution samples,measurement of glucose in serum environment is also demonstrated.Glucose samples of different concentrations are made with serum as the dissolvant,which is close to human body environment.Our spectral measurements indicated that serum dissolvant presented a stronger disturbance on the 3.4-3.7μm absorption feature as compared to the one at 8～10μm.The coefficients of determination（R~2）using the two absorption features are larger than 0.97for the measurement ranges of 0～1000 mg/dl and 0～400 mg/dl.The root mean square errors（RMSECV）are 9.5-16.6 mg/dl and 7.52-10.24 mg/dl for measurement ranges of0～1000mg/dl and 0～400mg/dl,respectively.The glucose detection performance by using 8～10μm absorption feature is better than that of using 3.4～3.7μm absorption feature.In addition,as compared with water glucose solution samples,the glucose detection performance becomes slightly worse due to the disturbance of serum environment.