Diffusion imaging of white matter fibre tracts

This thesis presents the design and validation of a method for digitally reconstructing white matter fibre tracts in vivo using magnetic resonance imaging (MRI). The technique uses diffusion weighted MRI to estimate a likelihood distribution function for the fibre direction(s) in each imaging voxel, and subsequently infers connectivity from any point in the central nervous system to another. The fibre tracking algorithm addresses issues that can confound fibre tract reconstruction, such as imaging noise, subvoxel partial volume averaging of fibre directions, and problems with the estimate of the diffusion probability density function (pdf). It can take as input a diffusion pdf estimated using either the traditional diffusion tensor approach or more recent high angular resolution diffusion approaches. The fibre tracking technique is validated using in vivo human brain diffusion imaging data and using a phantom constructed from excised rat spinal cord, which provides a "gold standard" connectivity map. The results are promising, especially for regions of the brain where tracking using previously described algorithms has been difficult to perform, for example, the regions of complex fibre structure near the cortex. As the cortex is critical for functional activity in the brain, this may have widespread implications for our understanding of the human brain in healthy subjects and in disease