Improved characterisation of crossing fibres: spherical deconvolution combined with Tikhonov regularization
Calamante F, Connelly A, Tournier J
Brain Research Institute
The diffusion tensor is commonly used to estimate fibre orientations from diffusion-weighted data, but cannot describe crossing fibre regions. We have recently proposed the spherical deconvolution technique for estimating the fibre orientation distribution (FOD) that effectively removes this limitation. Here we present a method for significantly improving the angular resolution of the reconstructed FOD, based on the unphysical nature of negative lobes in the FOD. This information was incorporated as a constraint in a Tikhonov regularisation formulation of spherical deconvolution. The improved angular resolution allows fibres crossing at smaller angles to be resolved, which is essential for reliable fibre-tracking.