Improved characterization of brain anisotropy using diffusion MRI
Second order diffusion tensor analysis of diffusion weighted MR data only accounts for a single intra voxel fibre direction. This poses a problem in many regions of the brain where fibres cross. An anisotropy measurement based on the traditional diffusion tensor model, such as fractional anisotropy (FA), produces significantly low values when there are fibres crossing within the same voxel, or in the presence of other partial volume effects. A new anisotropy index based on the variance of the diffusion MRI signal is described and applied to both simulated and experimental data. A method to normalise this parameter, in order to allow comparisons across scan sessions, is also presented. It is shown that this parameter can characterise white matter in situations in which the diffusion tensor formalism fails to accurately reflect the local diffusion. The images obtained show more detail in the fibre structure, a better contrast between regions of high and low anisotropy, and the main fibre tracts appear to be thicker and brighter, which corresponds better anatomically to the information obtained from structural images.