Section: New Results

Medical Image Computing in Brain Pathologies

Detection of cortical abnormalities in drug resistant epilepsy

Participants : Elise Bannier, Camille Maumet, Jean-Christophe Ferré, Jean-Yves Gauvrit, Christian Barillot.

Focal cortical dysplasia and heterotopias are a recognized cause of epilepsy. Indication for drug resistant epilepsy surgery relies on precise localization and delineation of the epileptogenic zone and lesion identification is an important issue. Visual detection and delineation of small or occult focal cortical dysplasia and heterotopias on MR images are sometimes difficult. The Double Inversion Recovery (DIR) imaging, by nulling white matter and cerebrospinal fluid signal, seems particularly appropriate to detect intracortical lesions in MS and Epilepsy. In this work we evaluated at 3T and using voxel-based morphometry (VBM) the ability of a 9-minute 3D DIR sequence to detect cortical and juxtacortical lesions in drug resistant epileptic patients. Results on 21 patients and 20 healthy volunteers show the potential of 3D DIR VBM to detect cortical abnormalities. Further work will investigate the use of alternate registration frameworks (e.g. DARTEL), improved intensity normalization of 3D DIR images and joint 3D T1-w/DIR analysis to improve detection sensitivity and specificity.

Multi-modal NMR cartography of USPIO positive and negative tissues in MS human models

Participants : Olivier Luong, Olivier Commowick, Christian Barillot.

The main objective of this work was to build an input object for an MRI simulator. Each voxel of the object is defined by its three physical entities which are T 1 , T 2 and ρ MR relaxation parameters. In our case, this object comes from Multiple Sclerosis brains. We initially defined a simplified model with respect to pathological regions, based on a combination of the Brainweb template and the lesion manually delineated from pathological images. From this, we allocated relaxation parameters for each voxels of these ROI based on fixed values of T 1 , T 2 and ρ (initialized from in vivo relaxometry acquisitions). This model model does not allow to obtain a fine description of the pathological regions, as potentially defned by differential contrats between USPIO and Gd enhanced images. In order to obtain this finer description, we used an MRI simulator based on the Bloch's equations, in order to estimated the T 1 , T 2 and ρ parameters on each voxel from initial conditions coming from in-vivo images acquired in Rennes by using the USPIO-6 protocol.

This work is part of the VIP collaborative project, supported by ANR (Agence National de la Recherche), through grant ANR-AA-PPPP-000. The partners with whom we have the tightest relations are: Creatis (Lyon), I3S (Sophia), CEA-LETI (Grenoble).