Meeting Banner
Abstract #4474

3D locally dependent regularization of the diffusion tensor using ICA and TGV

Gernot Reishofer 1 , Kristian Bredies 2 , Karl Koschutnig 3 , Margit Jehna 4 , Christian Langkammer 5 , David Porter 6 , and Hannes Deutschmann 4

1 Radiology, Medical University of Graz, Graz, Styria, Austria, 2 Institute for Mathematics and Scientific Computing, Universtiy of Graz, Austria, 3 Psychology, Universtiy of Graz, Austria, 4 Neuroradiology, Medical University of Graz, Austria, 5 Neurology, Medical University of Graz, Austria, 6 Siemens AG, Healthcare Sector, MR R&D, Germany

It has been shown recently, that spatially dependent regularization of the diffusion tensor applied on readout-segmented echo planar imaging (rs-EPI) with 2D navigator-based reacquisition significantly improves fractional anisotropy (FA) maps and tractography. In this work we propose a novel approach for automatic regularizing the entire diffusion tensor utilizing a three dimensional implementation of total generalized variation (TGV). The evaluation of the noise distribution of the diffusion tensor by means of ICA allows for an automatic update of the regularization parameter making the proposed algorithm user-independent. Furthermore the incorporation of the locally varying noise distribution allows for a spatially dependent regularization.

This abstract and the presentation materials are available to members only; a login is required.

Join Here