Probabilistic Models of Cortical Sulci :

Application to the Fusion of MEG and MRI Data




Competition of methods of inter-subjects fusion for the definition of probabilistic spaces adapted to the observation of specific cerebral activations

 This diagram clarifies the problems which we want to approach in this work. The first phase consists in collecting anatomical and functional information in IRM and MEG. Then, the "Local Registration" consists in segmenting the sulci of interests to model them in 3D. These sulci are then readjusted on the basis of a local referencial (i.e. a  reference related to the studied sulci) to allow a statistical analysis of the variations in form and position of these sulci. The statistical model is interpretable via its "mean"  representation and of its principal modes of deformation. Activations of the various subjects can be amalgamated in this average statistical space to deduce some from the probabilities of activation. This methodology can be then compared with the data processing sequence "Global Registration"

CONTEXT

One of the objectives of  functional neuroimaging is to carry out the cartography of cerebral functions. One of the major problems related to the establishment of these functional maps is the mapping of data collected from different subjects. This matching passes by the use of an anatomical substrate, generally by the means of Magnetic Resonance Imaging (MRI). MRI allows to obtain some 3D precise representations of cerebral anatomy and to visualize easily the cerebral cortex and its components gyri and sulci, place of (supposed) presence of many studied cognitive functions. Several arguments can argue on the interest of modeling cortical sulci in a process of anatomical and functional fusion:
 
  • Cortical sulci serve as landmarks associated to major functional regions.
  • Cortical sulci can be used as landmarks and so as a support for inter-individuals fusion processes.
  • Cortical sulci can help to the process of labelling cortical structures (gyri for instance).
  • Cortical sulci can be used as indices to some functional organizations.

    • OBJECTIVES

    This project is a contribution to the problem of statistical modeling of cortical sulci on a population of subjects, on the  numerical side (description of a analytical and statistical digital model describing sulci) as well as on the symbolic side (symbolic representation describing the nature of the sulci and the relations it shares with its anatomical and/or functional environment from one or more subjects). We propose on the one hand to show the interest of this statistical modeling for the interpretation of space-time localizations collected from several subjects using MEG. In addition, we propose to compare this approach of fusion using a local reference mark with approaches allowing the fusion of cerebral activations in a standardized space obtained after global matching (linear or non-linear) of the brains of the various studied subjects.

    PRIME CONTRACTOR

    Christian BARILLOT

    Ph: (+33/0) 2 99 84 75 05
    Fax: (+33/0) 2 99 84 71 71
    Email : Christian.Barillot@irisa.fr
     

    PROJECT PARTNERS

    VISTA Project (Main Contractor)

    IRISA - Institut de Recherche en Informatique et Systèmes Appliqués
    Unité de Recherche INRIA - CNRS de Rennes
    Campus Universitaire de Beaulieu
    35042 Rennes Cedex

    Contact : Christian Barillot

    Ph : (+33/0) 2 99 84 71 00  -  Fax : (+33/0) 2 99 84 71 71
    Email : Christian.Barillot@irisa.fr
    Http:  www.irisa.fr/vista
     

    SIM Laboratory (Signaux et Images en Médecine), UPRES EA 9606

    Faculté de Médecine
    Université de Rennes I
    35043 Rennes Cedex

    Contact : Bernard Gibaud

    Tel: [+33 2 / 02] 99 33 68 63  -  Fax: [+33 2 / 02] 99 33 68 88
    Email: Bernard.Gibaud@univ-rennes1.fr
    Http: sim3.univ-rennes1.fr
     

    EPIDAURE Project

    I.N.R.I.A. Sophia Antipolis
    2004, Route des Lucioles BP 93
    06902 Sophia Antipolis Cedex

    Contact : Gregoire.Malandain

    Tel: [+33 4 / 04] 93 65 79 26  -  Fax: [+33 4 / 04] 93 65 76 69
    Email: Gregoire.Malandain@sophia.inria.fr
    Http: www.inria.fr/epidaure/Epidaure-eng.html

    McConnell Brain Imaging Center

    Montreal Neurological Institute
    McGill University
    Montréal, Québec H3A 2B4, Canada

    Contact : Alan Evans

    Tel: (+1) (514) 398 1996  -  Fax: (+1) (514) 398 8948
    Email: alan.evans@bic-mni.mcgill.ca
    Http: www.bic.mni.mcgill.ca/