Unravelling the link between the structure and function of the human brain: A connectivity study using functional MRI and diffusion tensor imaging

Abstract

The brain structure-function relationship has been one of the most fundamental issues in neuroscience and the precise underpinnings of such a link has remained elusive. To address this, investigations have discovered associations between cortical structure or white matter architecture and function in the forms of brain activation, task performance, and/or cognitive dysfunctions. However, limited attention has been given to the search for a more direct link between task-related regions of cortical activation and how they are wired together. The overarching goal of this thesis is to gain a further understanding of this relationship by directly assessing functionally coupled regions and the underlying white matter architecture. The first study used diffusion tensor imaging (DTI) and tractography to provide a global outlook of structural asymmetries of white matter in the major lobes, and the extent to which they relate to known functional asymmetries. With the inclusion of the individual diffusion measures (mean diffusion [MD], parallel diffusion and perpendicular diffusion) in combination with fractional anisotropy (FA), the microstructure of the pathways generated from each lobe was closely examined. Structurally, asymmetries did indeed follow expected patterns of functional laterality in terms of language and visuospatial function. In order to validate this anatomo-functional relationship, studies two and three combined the methodologies of functional magnetic resonance imaging (fMRI) with DTI and tractography. In each individual, the regions that were shown to be functionally connected during particular cognitive tasks were identified. These sites were then used to track white matter pathways using tractography. As expected, study two showed a leftward functional network for language production and comprehension. A significant correlation between functional and structural measures was found for the fronto-motor pathway recruited during the generation of verbs. It is theorised that this pathway may be specific to verb generation, given previous reports of a possible motor component in the processing of verbs, while the longer-range connection to temporal regions may be involved in a more diverse range of functions than language alone. Interestingly, tractography between the regions involved in comprehension revealed a ventral as well as a dorsal white matter pathway, which is in accord with the previously proposed dual route theory of reading. Using the same approach, study three explored the more widely and bilaterally distributed circuitry of working memory to investigate whether a structure-function relationship could be established in more complex networks. To further extend the work on cerebral asymmetries, spatial and verbal elements were also included. Functionally, a typical fronto-parietal network was identified for both spatial and verbal working memory. In terms of functional and structural connectivity however, subtle yet significant differences were revealed within both networks that may reflect distinct cognitive processes required for each task. Functional connectivity showed a leftward pattern for the spatial task and rightward connectivity for verbal, contrary to early studies. Measures of white matter integrity on the other hand, were much more symmetrical across the tasks, though variations in the individual diffusion measures revealed possible differences in underlying axonal properties. However, the spatial working memory network almost exclusively yielded significant correlations between structural and functional measures in the parietal-premotor pathways, which may reflect the specificity of the pathways for spatial working memory. Furthermore, comparisons between the two tasks revealed increased activity in the left precentral gyrus during verbal working memory, consistent with the idea that spatial working memory involves a generic system for working memory, while verbal working memory requires additional recruitment of left hemispheric regions for processing linguistic stimuli.