Concentrations of blood constituents provide diagnostic information in clinical procedures. Among the medical community, there is great interest in developing an optical method of concentration measurements that eliminates reagents required in the current clinical chemistry techniques and is non-invasive. This thesis describes the methodology to develop a measurement technique of blood component concentrations using Raman spectroscopy. The sources of prediction error were identified. A Monte Carlo simulation model was developed to study the distribution of the Raman signal from turbid biological samples. A high sensitivity system for Raman spectroscopy of blood analytes was designed by optimizing the optical components. The new system had higher collection efficiency than the previous state-of-the-art system. Experiments were performed using the new system in order to collect Raman spectra of human whole blood. The collected Raman spectra were analyzed with multivariate calibration techniques, and compared with hospital measurements. The analysis of the spectra of whole blood samples demonstrate that many chemical components in ex vivo whole blood samples can be measured accurately with near-infrared Raman spectroscopy. A preliminary analysis based on the results of this thesis indicates that it is feasible to measure blood analytes non-invasively.
展开▼
