March 2002

BACK   

The talk will describe two new brain mapping methods designed to get closer to the class of neuroscience approaches dealing with structural models. The first approach aims at the inference of the cortex connectivity from the living human brain using MR diffusion imaging. The second approach is a structural alternative to the deformable brain atlas paradigm dedicated to the cortical shapes.

1. Brain connectivity:

Magnetic Resonance Diffusion Tensor Imaging (DTI) provides information about fiber local directions in brain white matter, which opens the door to the tracking of fiber bundles connecting different brain areas. Diffusion of water molecule, indeed, turns out to have a larger amplitude in the fiber direction. Unfortunately, diffusion information is difficult to interpret at the level of fiber crossing, which may lead to erroneous forks of the tracking process. I will present a regularization based solution consisting of minimizing the bundle curvatures. The reconstruction of the map of the main bundles is considered as an inverse problem, which solution is the minimum energy configuration of a simulated spin glass.

Reference:

• C. Poupon, C.A. Clark, V. Frouin, J. Régis, I. Bloch, D. LeBihan, and J.-F. Mangin.
  Regularization of diffusion-based direction maps for the tracking of brain white matter fascicles.
  NeuroImage, 12(2):184-195, August 2000.

2. Cortical folding patterns:

We describe a complete system allowing automatic recognition of the main sulci of the human cortex. This system relies on a complex preprocessing of MR images leading to abstract structural representations of the cortical folding. This preprocessing consists of a sequence of automatic algorithms mainly based on Mathematical Morphology. The representation nodes are cortical folds, which are given a sulcus name by a contextual pattern recognition method. This method can be interpreted as a graph matching approach, which is driven by the minimization of a global function made up of local potentials. Each potential is a measure of the likelihood of the labelling of a restricted area. This potential is given by a multi-layer perceptron trained on a learning database. A base of 26 brains manually labelled by a neuroanatomist is used to validate our approach. The system developed for the right hemisphere is made up of 265 neural networks. The whole system is a symbolic alternative to the usual deformable atlas principle. This alternative consists of using a higher level of representation of the data to overcome some of the difficulties induced by the complexity and the striking variability of the cortical folding.

Reference:

• D. Rivière, J.-F. Mangin, D. Papadopoulos, J.-M. Martinez, V. Frouin and J. Régis,
  Automatic recognition of cortical sulci using a congregation of neural networks,
  MICCAI'2000, Pittsburgh, LNCS-1935, Springer-Verlag, pp 40-49.

More Downloads:

• "From 3D Magnetic Resonance Images to Structural Representations of the Cortex Topography using Topology Preserving Deformations," JMIV, 1995.
• "A MRF Based Random Graph Modelling the Human Cortical Topography," CVRMed, pp. 177-183, 1995.
• "Robust brain segmentation using histogram scale-space analysis and mathematical morphology," MICCAI, pp.1230-1241, Lecture Notes in Computer Science, vol. 1496, Springer-Verlag, 1998.
• "Entropy Minimization for Automatic Correction of Intensity Nonuniformity," MMBIA, 2000.

BACK

Last Updated: Feb. 13, 2002