Accueil SDV
Autres sites CNRS

The NEMESIS group no longer exists. You can visit the website of our new team ARAMIS at http://www.aramislab.fr.




signal and image processing for neuroscience

Neuroimaging and electrophysiology (MRI, EEG, MEG), shape models, segmentation, neurodynamics, network modeling, statistical learning, applications to epilepsy, Alzheimer's disease, stroke



The NEMESIS group at the Research Center of the Brain and Spine Institute (CR-ICM) aims at designing new models and methods for images and signals of the healthy and pathological human brain.

Group leaders: Olivier Colliot (CR CNRS) and Didier Dormont (PU-PH)



Computational neuroanatomy

Computational anatomy, i.e. the computerized analysis of anatomical images such as MRI, opens fascinating perspectives for the understanding of neurological and psychiatric disorders. Being automatic and reproducible, it enables large-scale quantitative studies to discover the patterns of structural abnormalities associated to pathologies of the central nervous system.


Our team develops new methods for the quantitative study of brain structures. Our main research areas include :

  • Automatic segmentation of the hippocampus
  • Ultra-high field MRI (7T) [collaboration with Neurospin and University of Minnesota ]
  • Diffeomorphic cortical registration
  • MRI morphometry
  • High-dimensional image classification


In collaboration with the clinical departements of La Pitié-Salpêtrière hospital, these methods are applied to the study of different neurological and psychiatric disorders, including Alzheimer's disease (collaboration with Stéphane Lehéricy and Bruno Dubois teams), depression (collaboration with Philippe Fossati team), epilepsy and stroke.


For more informationFor more information, please contact


SACHA: automatic segmentation of the hippocampus and the amygdala from MRI


DISCO: diffeomorphic sulci-based cortical registration




Dynamic brain networks

Many brain processes, such as cognitive tasks and epileptic phenomena, rely on synchronization of distant neural assemblies. These processes are of high complexity and emerge at different scales ranging from the single cell to large brain areas. To that end, it is necessary to develop new tools well adapted to complex properties of neural signals.


The overall goal of our research is to develop mathematical approaches to model the dynamic and topological properties of brain networks, in order to properly characterize neural interactions and to study how local interactions (at a cellular level) scale-up to a global coherent dynamics in the brain during normal and pathological brain states.


For more informationFor more information, please contact




Multimodal imaging of Epilepsy

Our research is focused on refractory epileptic patients candidates to surgery for better delineating the epileptic focus and predicting and avoiding cognitive defects after surgery.


For the first objective (better delineation of the epileptic focus), we wish to test the localizing value of multimodal neuroimaging by coupling and evaluating morphometric MRI with new segmentation methods, diffusion MRI techniques, spectroscopy MRI, nuclear imaging including SPECT and PET and EEG-MEG.


For the second objective, we wish to determine which of BOLD functional MR imaging (fMRI) or Wada test (the intracarotid sodium amobarbital procedure) are the best predictors of postoperative memory changes in epileptic patients who will undergo temporal lobe surgery and whether fMRI may replace the Wada test.


For more informationFor more information, please contact

frmi hippocampus

Left hippocampal activation using fMRI




Neuroimaging of Stroke

Stroke is of major interest in clinical research since it is the first cause of motor disability, the second cause of dementia and the third cause of mortality. Stroke research in the COGIMAGE group is based on two different endpoints, using Magnetic Resonance Imaging (MRI).


  • The first axis consists in understanding the mechanisms of recovery after sroke. Brain plasticity after stroke relies on cortical reorganization and white matter pathways. We propose to study these phenomenons with functional MRI and tensor diffusion imaging.


  • The second one is focused on disability prediction. Improving prediction of stroke outcome at the acute stage would assist therapeutic decision making. An in-house image analysis software (Neurinfarct) has been developed to predict infarct volume at day one using DWI (diffusion weighted imaging). Early prediction of disability is also a major goal and MRI-tools could help to predict stroke prognosis.


For more informationFor more information, please contact


NEURiNFARCT: software to predict infarct growth in acute stroke









Sur le WEB
du laboratoire

Sur le WEB du CNRS

Imprimer Contact Plan du site Crédits Plug-ins Accueil