Blanc SIMI 1 - Sciences de l'information, de la matière et de l'ingénierie : Mathématiques et interactions

Combining large dimension data analysis for neuroimaging and genetic – GENIM

Submission summary

We propose to develop methods to analyse conjointly phenotypes from behavioural and neuroimaging data, and genetic data (Single Nucleotype Polymorphism) of large dimension. Neuroimaging should act as an intermediate endophenotype and help to understand the link between genetic and phenotypic variability. This would permit advances both in the clinical domain (towards individualized treatment) and in the study of brain functions in normal subjects.

This proposal gathers expert partners in neuroimaging data analysis (Neurospin), applied statistics (Supelec, Neurospin), cognitive neuroscience (Inserm U562) and human genetics (Pasteur).
Our participation in the European project IMAGEN will allow us to work with data acquired on a cohort of adolescents in whom risk factors for addiction will be studied. On this cohort, neuroimaging data are being acquired (T1, BOLD, and diffusion data) from eight neuroimaging centres across Europe. One thousand subjects have already been assessed for neuroimaging, and 750 have been genotyped, yielding 650 000 SNP measures on the genome. These measurements are complemented with more than 20 behavioural / neuropsychological tests. Neurospin is in charge of constructing the central database grouping those data for the European consortium. Moreover, U562 has acquired fMRI, T1 and 1M SNP data for the "Localizer" that includes more than 100 subjects.

These recent data are challenging in terms of their analysis. Data that couple both neuroimaging and genetic information are of very high dimension with respect to the size of the cohort of subjects, and classical hypothesis testing statistical techniques are not adapted and will lack power. Besides, neuroimaging and genetic data must also be combined with behavioural or demographic variables. We intend to explore several axes of research avenues for the analysis of these datasets (Neurospin, Supélec), and also to complete and extend the "Localizer" dataset with a new protocol using experimental conditions that reflect the social interaction processes (the "social localizer", Inserm U562 and Pasteur).

We will first attempt to reduce the dimension of neuroimaging datasets, in particular on fMRI data, with techniques adapted to inter individual variability. In parallel, we will develop and test new data reduction strategies on genetic datasets (SNP with or without information on haplotypes) by means of multivariate methods, and we will evaluate multiple comparison techniques. A third task will be to develop, test and validate analysis techniques based on regularized and generalized canonical correlation analyses, whose principle is to extract exploratory variables from each block (or type) of data that show a relation with explanatory variables from other blocks. We will extend these techniques and propose kernel and sparse versions. We will validate from a methodological stand point using resampling techniques (cross-validation). An important aspect will be the interpretation of the results by partners in cognitive neuroscience (U52) or genetics (Pasteur).

Software development will hold an important place in the project. They will be organised such that efficient and documented methods will be available for use by our partners in a first instance, and then by the community at large.
We hope that this project will bring novel and innovative techniques to explore and exploit data for which there is yet no adapted method. These analysis techniques should help to better understand the link between neuroimaging endophenotype and genetic variability across subjects.

Project coordination


The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.


Supélec Département Signaux & Systèmes Électroniques, Ecole supérieure d'électricité
Pasteur Unté de génétique humaine et fonctions cognitives, Pasteur
Inserm Unicog U562 Inserm

Help of the ANR 110,000 euros
Beginning and duration of the scientific project: - 36 Months

Useful links

Explorez notre base de projets financés



ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter