Computer vision and machine learning models of brain imaging variability: from population-wide databases to precision models for high-field MRI
The GAIA laboratory is part of the BAOBAB (CEA, CNRS, Paris-Saclay University) unit in the NeuroSpin departement. Directions to Neurospin.
Leader:
Develops innovative methods to decipher the cortical folding pattern and perform automatic sulcus recognition, based on the use of machine learning and artificial intelligence techniques to infer dictionaries of tractable local polymorphisms supposed to provide a good proxy to the architectural segregation of populations.
Leader:
Develops innovative analysis methods to integrate brain imaging with genomics data, often combined to clinical or environmental data to determine how brain phenotypes variability originates in genetic and environmental variabilities.
Leader:
Develops innovative methods to map the brain structural connectome, cortex cytoarchitectony and myeloarchitectony at the individual scale, from animals to humans using ultra-high field and extreme field MRI.
Leader:
Develops cutting-edge machine learning and statistical models to uncover neural signatures that can stratify psychiatric disorders in the frame of large-scale national and European initiatives.
Leader:
Founded in 2010, CATI, or Centre d'acquisition et de traitement des images in French, is a national service platform supporting more than 30 muticenter neuroimaging studies in the fields of neurodegenerative diseases and psychiatry (AD, Parkinson, Huntington, ALS, Bipolar, etc.) including several therapeutic trials.
Leader:
Our goal is to meet the specific needs of NeuroSpin researchers for software development and analysis of heterogeneous brain data (neuroimaging, omics, clinical and behavioural data) used in the context of research conducted in clinical neuroscience, population imaging or cognitive neuroscience research.
Exploring the potential of representation and transfer learning for anatomical neuroimaging - application to psychiatry.
Surfify is an open source Python module that simplifies the development of neural network architectures that relies on cortical surfaces. It provides common architectures, icosahedral mesh operators, and cortical augmentations.
White matter atlases of the human [a], chimpanzee [b] and Japanese quail brains [c]; Medusa simulations of realistic cortex [d] and white matter [e] brain tissues; color-encoded direction map of the Chenonceau brain scanned using 11.7 Tesla diffusion MRI at the mesoscale [f] and inference of the fine structural connectivity of the Chenonceau brain cortical ribbon using HPC spin- glass tractography [g].
BHB aggregates 10 publicly available datasets. Currently, it is focused only on Healthy Controls (HC) since the main challenge consists in modeling the (normal) brain development by building a robust brain age predictor. BHB contains N=5330 3D T1 brain MRI scans from HC acquired on 71 different acquisition sites coming from European-American, European, and Asian individuals. BHB provides the participants phenotype as well as site and scanner information associated with each image. Some widespread confounds are also proposed, such as the Total Intracranial Volume (TIV), the CerebroSpinal Fluid Volume (CSFV), the Gray Matter Volume (GMV), and the White Matter Volume (WMV).
Using data from the UK Biobank, the world's largest general population imaging-genetics cohort, a collaboration led by a team at BAOBAB (NeuroSpin) has identified genes involved in the genetic architecture of functional language connectivity. These genes could be a priority for studying natural language.
We examined here the morphological variability of the central sulcus in hominids, with a focus on the motor hand region, significant issue of human evolution.