Functional connectivity captures brain interactions via fluctuations in the observed activity. We use graphical modelling to extract connectomes, i.e. graphs of brain connectivity, with statistically well-posed and controled methods.
Learning brain connectivity faces the complexity of brain graphs, that is in sharp contrast with the small amount of data at the subject level. At the group level, connectivity information can be convolved with inter-subject variability. To tackle these challenges, our research adapts cutting-edge methods from the statistical learning literature to the specificities of brain imaging.
- Markov models for fMRI correlation structure: is brain functional connectivity small world, or decomposable into networks? G Varoquaux, A Gramfort, J-B Poline, B Thirion, Journal of Physiology – Paris, 2012
- Brain covariance selection: better individual functional connectivity models using population prior G Varoquaux, A Gramfort, J-B Poline, B Thirion. Advances in Neural Information Processing Systems, 2010
- A Novel Sparse Graphical Approach for Multimodal Brain Connectivity Inference B Ng, G Varoquaux, J-B Poline, B Thirion, Medical Image Computing and Computer Assisted Intervention, 2012
Inter-subject bio-markers from connectivity require statistical modelling of the distributed variability in the connectomes. Univariate statistics are ill-suited for these problems, and we develop matrix-variate probilistic models that lead to well-posed statistical tests maximizing the edge-level detection power.
- Detection of brain functional-connectivity difference in post-stroke patients using group-level covariance modeling G Varoquaux, F Baronnet, A Kleinschmidt, P Fillard, B Thirion, Medical Image Computing and Computer Added Intervention, 2010
- Hyperfrontality and hypoconnectivity during refreshing in schizophrenia M-L Grillon, C Oppenheim, G Varoquaux, F Charbonneau; A-D Devauchelle; M-O Krebs, F Bayle, B Thirion, C Huron Psychiatry Research: Neuroimaging, 2013