Microstructure imaging with diffusion MRI and microscopy

Thesis

Diffusion MRI is a powerful, non-invasive tool for microstructure imaging because of its sensitivity to structures on the microscopic scale. By analysing the diffusion signal we can infer the structural connectivity of the human brain or changes in the myelo- and cyto-architecture. This inference requires biophysical signal modelling, where we aim to map macroscopic diffusion signals onto biologically-meaningful tissue parameters. However, this mapping is often ill-posed, with multiple parameter sets fitting the signal equally well, and our parameters are said to be degenerate. Because degenerate parameters lack tissue specificity and are thus difficult to interpret, they present a serious challenge to diffusion modelling and limit the widespread application of diffusion MRI in both neuroscience research and clinical medicine.

This thesis explores various ways in which we can overcome parameter degeneracies to enable more specific microstructural interpretation of the diffusion signal. Consequently, this work spans species (human and macaque), tissue states (in vivo and postmortem), modalities (MRI and microscopy) and signals over four orders of magnitude, from sub-micron microscopy images to MRI at the millimetre scale. In particular, we develop two biophysical models of the white matter that aim to separate the diffusion characteristics from the tissue architecture (i.e. the fibre orientation, dispersion and density) and present a truly unique, multimodal dataset: the BigMac dataset. This dataset combines extensive MRI data with multi-contrast microscopy to characterise a single macaque brain in exquisite detail and will establish an innovative data platform from which we can interconnect microstructural features with MRI signals throughout the brain.

Overall, the models and data presented in this thesis aim to further our understanding of how microscopic tissue features influence the observed diffusion signal. We hope that this work will enhance the biophysical interpretation of diffusion MRI and ultimately advance in vivo quantification of the brain tissue microstructure.

Authors: Howard, A

• DOI: 10.5287/ora-jnwvdp0gr
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: microscopy; data fusion; diffusion modelling; polarised light imaging; histology; brain imaging; diffusion MRI; microstructure Imaging; white matter
• Subjects: Microscopy; Diffusion magnetic resonance imaging