Claudia A.M. Wheeler-Kingshott1, Daniel C. Alexander2, Torben Schneider1, Mara Cercignani3
1Department of Neuroinflammation, UCL Institute of Neurology, London, UK; 2Dept. Computer Science, University College London, Centre for Medical Image Computing, London, UK; 3Neuroimaging Laboratory , Fondazione Santa Lucia, Rome, Italy
The diffusion coefficient along and across white matter fibers is a useful biomarker for tissue changes, such as axonal loss and demyelination. Axial and radial diffusivities have been associated with the DT eigenvalues, but crossing fibers, pathology and noise can affect the orientation of the corresponding eigenvectors. Here we present a method for calculating the axial and radial diffusivities in the brain based on the directionality of the structures in a healthy super-dataset representing the average diffusion properties of the population under sample. The use of this super-dataset helps overcoming the issues of pathology, noise and potentially of crossing fibers.