Study Of Theories And Experiments On Urine Components Detection By Multi-spectroscopic Method

Spectroscopy analysis technique is efficient, rapid, no reagents and can be run at low cost. It has become an important method in clinical analyses field. The success of using spectroscopy method in urine quantitative analysis will have a profound affection on the clinical biomedical analyses. Urine contains a wide variety of substances. The change of components in urine can not only reflect pathological change in urinary system and reproductive system, but also relate to a lot of whole body diseases. In the dissertation, Near-infrared (NIR) , Mid-infrared (MIR) and Raman spectroscopy method were used in the quantitative measuring of important components in urine. The optimal predict condition was select for different spectroscopy method, and the spectroscopic method successfully accounts for urine components variation.The main research content of the dissertation involves:1. A systematic theoretical and experimental research work about using near-IR, mid-IR and Raman spectral methods for the determination of important components in urine were researched. This research integrated applied the advantage of three different spectral method in the urine components detection, and the present work can quantitative analysis main ingredients in urine and its spectral characteristics. In terms of spectral output, the largest difference between near-IR and the other two spectroscopies are in the degree of peak broadness and spectral overlap. Experiments have shown that mid-IR has slightly better predict ability in urine albumin than near-IR. At the same time, result showed that near-IR was superior in determination of urine glucose comparing to the mid-IR.2. The optimal conditions for urine albumin and glucose measurement were put forward. For near-IR, the optimal optical pathlength is 1mm, the model band for glucose and albumin is 4400～4700 cm^-1 and 4250～4850 cm^-1 respectively. After the application of genetic algorithms used for wavelength selection, Partial least- squares regression was used to model the concentrations of glucose and albumin, the RMSEP is 20.61 mg/dL and 20.95 mg/dL respectively. ATR method was used in mid-IR measurement, model region for glucose and albumin are 1200-950 cm^-1 and 1400-1700 cm^-1 respectively, genetic algorithm was used for wavelength selection, the Partial least- squares regression result for glucose and albumin are 28.57 mg/dL and 17.89 mg/dL respectively. For both near-IR and mid-IR, the detection error of glucose and albumin can meet the clinical application level.