Stephan Meesters1, Gonzalo Sanguinetti1, Eleftherios Garyfallidis2, Jorg Portegies1, and Remco Duits1
1Department of Mathematics and Computer Science, Eindhoven University of Technology, Eindhoven, Netherlands, 2Computer Science Department, University of Sherbrooke, Sherbrooke, QC, Canada
We present a novel open-source module that implements a
contextual PDE framework for processing HARDI data. It’s potential in enhancement
of ODF/FOD fields is demonstrated where the aim is to enhance the alignment of elongated
structures while preserving crossings. The method for contextual enhancement is
based on a hypo-elliptic PDE defined in the domain of coupled positions and
orientations and can be solved with a shift-twist convolution. The module is available
in the DIPY (Diffusion Imaging in Python) software library, which makes it
widely available for the neuroimaging community.
