Reconstruction of Fiber Orientation Distributions and Fiber Densities with Global Prior Knowledge based on Diffusionweighted MR-Imaging

• Applicant: Dr. Marco Reisert
• Subject Area: Human Cognitive and Systems Neuroscience
• Term: from 2013 to 2016
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 232066167

The analysis and interpretation of diffusion-weighted magnetic resonance imaging data is of great interest in neuroscience. The ultimate goal is the representation of the nerve fiber bundles in the human brain to understand the underlying connectivity relations. It is known that diffusion tensor imaging is not sufficient to describe the complex fiber geometry. Recently developed alternatives consider complete fiber orientation distributions rather than tensors. The study of methods for reconstructing such distributions is the central task of this project. Since the reconstruction problem is inherently ill posed, the inclusion of prior knowledge is necessary. Existing methods use here mainly local knowledge, i.e. the reconstruction of the fiber orientation distributions per volume element is independent. Assuming that the measured tissue is of fibrous nature, it is possible to formulate continuity conditions, which establish relations between adjacent volume elements in a global way. In this way, the estimate of the fiber orientation distributions can be made much more robust, which, on the one hand, reduces the demands on the quality of the measurement and, thus, the burden to the patient. On the other hand, new modalities can be derived, which are not available on the basis of local knowledge only. For example, it is possible to overcome the limitation caused by the measurement of centrally symmetric distributions. Together with boundary conditions indicators for the absolute number of fiber bundles can be derived.

• DFG Programme: Research Grants