Development of a Near-infrared Detection System for Oxidative Stress Analysis of Pregnant and Non-pregnant Mouse Serum Samples with Parasite Loading.

摘要

This Thesis presents a quantitative tool that has been developed for the oxidative stress analysis of mouse serum using near-infrared (NIR) spectroscopy. We hypothesize that elevations in oxidative stress will cause changes in pre-selected NIR spectral regions. An optical detection system has been developed and optimized using bovine serum albumin. A total of 118 mouse serum samples were collected and NIR spectra were obtained. Sample collection was carried out in two stages. First, 70 mouse serum samples were collected followed by an additional 48 samples. A pre-selection of 5 hypothesis-driven wavelength regions (CH, SH, POH, ROH, and RNH2) were used in order to obtain spectral signatures for the serum samples. Spectra were analyzed using multi-linear regression (MLR) based optimizations were all possible variable combinations and ratios were computed. The Students' t-test for statistical significance was used to obtain the most parsimonious combination of variables.;This proposed research offers the basis for a rapid point-of-care method that could be used to differentiate samples with infection.;Pregnancy and infection status were classified using this approach. Mouse serum samples with doses of a murine gastrointestinal nematode were investigated. NIR spectral differences were observed for pregnant and non-pregnant serum samples with varying infection states in the 1600-2400 nm spectral window. Of the 5 pre-selected spectral regions, results indicate that 3 components (CH, SH, and POH) demonstrate the most pronounced changes in their relative means for non-pregnant mouse serum samples compared with non-infected and high infection states. The proposed 3-component model was used to differentiate between non-infected and high infected mouse serum samples with >95% confidence (sensitivity=82% and specificity=81%). Additionally, the proposed 3-component model was used to determine if a low infection state could be differentiated from both non-infected and high-infected samples. Significant differences with >95% confidence were obtained for mouse serum samples which were non-infected compared to those exhibiting high-infection states. These findings are significant in that they suggest an intermediate infection level may be both present and also quantified. The system was then also used to differentiate between infection states in 48 pregnant serum samples. The equation generated by the MLR was solved for the regions yielding the highest separations. Results showed improved separations with >95% confidence (sensitivity=86% and specificity=81%). The results confirm the hypothesis that CH, SH, and POH functional groups may be used as markers for the progression of infection. It should be highlighted that separations between infection states were achieved with >95% confidence independent of pregnancy status.