Research On The Technology Of Noninvasive Blood Glucose Near-infrared Spectroscopy Signal Detection And Processing

At present,the global increase in the incidence of diabetes has become one of the most serious public health problems in the world.Diabetes treatment requires the frequent collection of blood samples to monitor changes in blood glucose levels,which can cause physical and psychological distress to the patient and increase the risk of infection.Noninvasive blood glucose testing has the advantages of non-invasive,environmentally friendly and fast,can effectively alleviate the pain of patients in the blood glucose testing,and can achieve continuous measurement,is the development trend of blood glucose testing.Among the noninvasive detection techniques of blood glucose,NIRS has the advantages of convenience,low cost,and fast analysis.It is one of the most widely researched and mature methods for noninvasive detection in the human,which has good potential and development prospects.In this thesis,the detection and processing technology of noninvasive blood glucose near-infrared spectroscopy signal is studied.A noninvasive blood glucose detection system based on near-infrared spectroscopy is designed.Experiments using different wavelengths,based on partial least squares regression method to establish a glucose correction model,and select the best measurement wavelength.The main work and achievements of the dissertation are as follows:1.The theoretical basis and characteristics of near-infrared spectroscopy technology were studied.By analyzing the near-infrared absorption spectra of water and glucose and the near-infrared absorption peaks of the main components in the blood,a suitable detection wavelength range of 1500 nm to 1700 nm is obtained.By analyzing near-infrared light after irradiation to human skin propagation path,select the finger as a test location.2.The near-infrared spectroscopy quantitative analysis method and the establishment for calibration model of non-invasive blood glucose detection were studied.The multiple regression modeling methods such as multiple linear regression,principal component regression,and partial least squares regression were analyzed,and partial least squares regression was chosen as the modeling algorithm of this study.3.A noninvasive blood glucose detection system based on near-infrared spectroscopy was designed.The system includes tunable laser,fiber optic probe,photodetector,data acquisition and processing modules.The system can choose different wavelengths of near-infrared light,at the same time,the introduction of blood pressure,heart rate and other parameters to reduce the impact of individual factors on the system.4.The blood glucose correction models were established.The experiments were carried out at 1530nm,1550nm,1570nm,1590nm and 1610nm wavelength respectively.Partial least squares regression models are built between spectral data,blood pressure,heart rate and blood glucose concentration,and get blood glucose correction models at different wavelengths.By analyzing and comparing the performances of these models,the optimal measurement wavelength is obtained at 1610 nm.The correlation coefficient of the corresponding blood glucose calibration model was 0.8460,the root mean square error of calibration was 0.5476 mmol/L,the root mean square error of prediction was 0.5557 mmol/L,which could predict the blood glucose concentration well.