Research On Influence Of Instrument Parameters To Noninvasive Biomedical Measurement By Near-infrared Spectroscopy

Biochemical examination of blood is one principle approach to diagnosis people’s health. Hyperglycemic and hyperlipemia threaten human health seriously. Continuous monitoring biochemical indexes of people with hyperglycemic and hyperlipemia is helpful to take effective steps and achieve precise control. Home blood glucose meter are currently invasive or minimally invasive, while there is no portable instrument for blood fat measurements. It is difficult to achieve effective disease prevention and precise control, due to lack of continuous real-time monitoring for blood glucose and fat. As a fast and comfortable examination approach, noninvasive biochemical examination is developing direction of biochemical examination. Near infrared (NIR) spectroscopy analysis is a main approach to achieve noninvasive biochemical examination.Noninvasive blood glucose monitoring is the main representative noninvasive biochemical examination, with a significant social impact. According results of previous studies, the currently key challenges that NIR noninvasive glucose detection faced are: 1) weak effective signal; 2) interference of tissue background; 3) blood volume changes. Our group has proposed NIR subtracted blood volume spectrometry, which can solve challenges 2) and 3). In this paper we discuss how to solve challenge 1).In order to solve the problem that the signal is weak, existing common spectrometers were compared, and performances were tested. They can not fully meet the needs of our non-invasive tests. We proposed to design a near infrared spectroscopy system, which can acquire information at multi-wavelength simultaneously.Influences of some instrument parameters to noninvasive biochemical examination were studied. 1) Spectrometer performance which biochemical examination required and performance of existing equipment were compared. 2) Based on experiments, correspondence between the sample thickness and appropriate near-infrared band was proposed. In serum analysis, the appropriate thickness of the sample and spectral region was determined. 3) Based on physical absorption characteristic, a method was presented to select spectral width for modeling. This method reduces spectral range for model calibration, helping improve the spectral signal to noise ratio and model stability, and it has been applied in prediction model establishment of serum cholesterol and triglycerides. 4) Using the same appropriately wide spectra, models were calibrated by some algorithms separately. Comparing performance of those models, a concise and robust algorithm was chosen to calibrate models for serum cholesterol and triglyceride. This procedure can be popularized to other analytes. 5) We simulated analyzed performances with different instrument bandpasses. During biomedical analyzer designing, the results provide reference to select instrument parameters. 6) After comparing component and pre-experiment, a spectrum collecting system was built.In the practical conditions, the system’s signal to noise ratio of reproducibility is greater than 10000:1.