ENHANCING NEUROIMAGING STUDIES AT MULTIPLE SCALES WITH STRUCTURAL MRI AND MULTIMODAL INFERENCE TECHNIQUES

Abstract

Advances in multimodal MRI have enabled scientists to probe neurological phenomena across scales. By characterizing the brain at the molecular, cellular, and macroscopic levels, multimodal MRI allows complementary information about the brain to be captured and analyzed together, producing more comprehensive understandings of neurological processes. However, each modality has its own weaknesses. Thus, the pursuit of characterizing the brain with multimodal MRI comes with the key challenge of overcoming, rather than compounding, each modality’s respective flaws. In this dissertation, we present several studies to address this challenge. Specifically, we propose new methodologies to improve our ability to capture and understand neurological phenomena across multiple scales with MR spectroscopy (MRS) and functional MRI (fMRI) and diffusion MRI (dMRI) by leveraging structural MRI and machine learning. First, we focus on innovations regarding the localization of spatially ambiguous molecular and functional signals from MRS and fMRI, respectively, through the superior spatial resolution and accuracy provided by structural MRI. Then, we move from the subcellular to the cellular scale. Driven by the challenge of being unable to conduct tractography-based analyses when complex microstructural dMRI acquisitions are unavailable, we investigate how anatomical context derived from structural MRI approximates the tractography operator and thus may facilitate these analyses more broadly. Finally, moving to the macrostructural, we focus on how intelligently supplementing macrostructural diffusion information with structural information allows for the generation of novel hypotheses regarding the underlying neuroanatomical processes in Alzheimer’s disease. Taken together, the contributions of this dissertation expand on the utility and capabilities of MRI modalities at multiple scales and contribute novel methodologies to the intersection of multimodal MRI and neurology.