A nonparametric reconstruction and its matrix implementation for the diffusion orientation transform (DOT)

The diffusion orientation transform (DOT) enables the computation of orientational probability profiles from high angular resolution diffusion-weighted magnetic resonance imaging data, making it possible to map connectivity information even in regions with complex microstructure. The initial formulation of the DOT has yielded probabilities in terms of spherical harmonics. In this paper, an alternative reconstruction method is introduced that makes it possible to directly compute the probabilities. The proposed reconstruction effectively combines the reconstruction and surface estimation steps in the visualization of the probability profiles. A matrix formulation is proposed that simplifies the implementation of the DOT method. The simulation as well as experimental results demonstrate the accuracy and robustness of the approach