Back to overview

Brain connectivity analysis in High Angular Resolution Diffusion Imaging

English title Brain connectivity analysis in High Angular Resolution Diffusion Imaging
Applicant Thiran Jean-Philippe
Number 103595
Funding scheme Project funding (Div. I-III)
Research institution Laboratoire de traitement des signaux 5 EPFL - STI - IEL - LTS5
Institution of higher education EPF Lausanne - EPFL
Main discipline Information Technology
Start/End 01.04.2004 - 31.03.2006
Approved amount 97'278.00
Show all

All Disciplines (2)

Information Technology

Keywords (9)


Lay Summary (English)

Lay summary
This project is the continuation of a previous Swiss NSF project onGeometrical Flow theory (active contours/surfaces) on non-scalar images,with application to brain connectivity analysis by Diffusion TensorMagnetic Resonance Imaging (DT-MRI). This recent medical imaging modalityallows obtaining non-invasive information about the ability of watermolecules to diffuse around their equilibrium position in a givendirection at every point (voxel) of an organ like the brain. As thisability to diffuse is closely related to the orientation of the neuralfibers in the brain, by processing this diffusion information, it ispossible to infer the neural connectivity between different regions in thebrain as well as to segment the major fiber tracts in vivo.

The basic assumption of the DT-MRI model is that the diffusion follows aGaussian probability distribution function. This hypothesis isquestionable in many situations in the brain. Indeed non-Gaussiandiffusion can arise from restricted diffusion or from slow exchangebetween partial volume compartments of Gaussian diffusion. The real tissuestructure is also a lot more complicated that can be described by asimplified model like the tensor model. For example fiber crossings cannot be represented by such a model. Therefore more sophisticated ways haveto be found to capture the structural information at each voxel usingDiffusion MRI. This basically means that we need to capture the diffusionproperty of the tissue not only following the Gaussian model assumed inDT-MRI, but by capturing the diffusion with high angular resolutiondiffusion (HARD) imaging. Techniques have been proposed in the literatureto tackle this problem. However, they are generally impossible to be usedin a clinical setting using whole body MR scanners. Their complexity islinked with the fact that they capture too much information, as comparedto what is needed for assessing the diffusion profile as we want to do inthis project.

Therefore the two problems that we want to address in this project are thefollowing ones:
- How can we measure High Angular Resolution Diffusion (HARD) probabilitydistribution functions (PDF) using a realistic MR acquisitionprotocols?
- How can we extract usable anisotropy information and structurecharacteristics from this HARD data, to be included in a Geometric Flowsegmentation model?

For the first problem, we propose to develop algorithms based on theso-called Persistent Angular Structure (PAS) that aims at determining aprobability density function with only angular dependence from diffusionMRI.

Then, the HARD information will have to be properly decomposed to beincluded in a non-scalar image segmentation framework using the activecontour theory, thank to the experience gained in the previous NSFproject.

At the end of this project, we will thus have a efficient way to measurethe diffusion information in every voxel of a brain, with high angularresolution, as well as segmentation algorithms to exploit thisinformation, that will be used for a better understanding of the brainconnectivity.
Direct link to Lay Summary Last update: 21.02.2013

Responsible applicant and co-applicants


Name Institute

Associated projects

Number Title Start Funding scheme
66943 Geometrical restoration and tracking in non-scalar images with application to Diffusion-Weighted Magnetic Resonance Imaging 01.04.2002 Project funding (Div. I-III)
121945 New methods for mapping and analysing large scale structural brain connectivity with MRI 01.08.2009 Project funding (Div. I-III)