Martin David King1, Chris Alan Clark1, David Gadian1
1UCL Institute of Child Health, University College London, London, UK
A spatial random effects model has been used in the analysis of multiple-directions diffusion-weighted MR data, focussing on the crossing-fibre problem that arises in tractography. The analysis was performed with the Markov chain Monte Carlo Gibbs sampler algorithm, using crossing fibre low b-value (1000 s mm-1) 20 directions diffusion data acquired from the pons. The results show that spatial random effects modelling, based on a conditional autoregressive prior, provides well resolved components compared with those obtained with an equivalent independent voxel-by-voxel treatment. We conclude that the spatial random effects modelling approach provides a useful extension to current tractography methods.