abstract: Hypoxia is a pathophysiological condition which results from lack of oxygen supply in tumors. The assessment of tumor hypoxia and its response to therapies can provide guidelines for optimization and personalization of therapeutic protocols for better treatment. Previous research has shown the difficulty in measuring hypoxia anatomically due to its heterogenous nature. This makes the study of hypoxia through various imaging modalities and mapping techniques crucial. The potential of hypoxia targeting T1 contrast agent GdDO3NI in generating hypoxia maps has been studied earlier. In this work, the similarities between hypoxia maps generated by MRI using GdDO3NI and pimonidazole based immunohistochemistry (IHC) in non-small cell lung carcinoma bearing mice have been studied. Six NCI-H1975 tumor-bearing mice were studied. All animal studies were approved by Arizona State University’s Institute of Animal Care and Use Committee (IACUC). Post co-injection of GdDO3NI and pimonidazole, T1 weighted 3D gradient echo MR images were acquired. For ex-vivo analysis of hypoxia, 30 μm thick tumor sections were obtained for each harvested tumor and were stained for pimonidazole and counter-stained with DAPI for nuclear staining. Pimonidazole (PIMO) is clinically used as a “gold standard” hypoxia marker. The key process involved stacking and iterative registration based on quality metric SSIM (Structural Similarity) Index of DAPI stained images of 5 consecutive tumor sections to produce a 3D volume stack of 150 μm thickness. Information from the 3D volume is combined to produce one final slide by averaging. The same registration transform was applied to stack the pimonidazole images which were previously thresholded to highlight hypoxic regions. The registered IHC stack was then co-registered with a single thresholded T1 weighted gradient echo MRI slice of the same location (~156 μm thick) using an elastic B-splines transform. The same transform was applied to achieve the co-registration of pimonidazole and MR percentage enhancement image. Image similarity index after the co-registration was found to be greater than 0.5 for 5 of the animals suggesting good correlation. R2 values were calculated for both hypoxic regions as well as tumor boundaries. All the tumors showed a high boundary correlation value of R2 greater than 0.8. Half of the animals showed high R2 values greater than 0.5 for hypoxic fractions. The RMSE values for the co-registration of all the animals were found to be low further suggesting better correspondence and validating the MR based hypoxia imaging. / Dissertation/Thesis / Masters Thesis Biomedical Engineering 2018