Formation / Training course
"Signal sur graphes; application aux neurosciences / Graph signal processing; application to neurosciences"
3-4 Apr. 2017
Institute for Language, Communication, and the Brain

Inscription / Registration

L'inscription est gratuite mais OBLIGATOIRE. Les inscriptions sont closes. The registration is free but MANDATORY. Registration closed. .


CH7 (Building 7, ground floor)
Aix-Marseille Université
La formation aura lieu sur le campus St Charles Bâtiment 7, salle CH7 (rez de chaussée) (même Bâtiment que la FRUMAM mais au -1).
The training will take place in building 7 room CH7 (same building as FRUMAM but ground floor (-1)) on the campus of Aix-Marseille University.

Téléchargement / Download

Merci de télécharger les supports nécessaires à la formation ICI.
Please download the necessary training material HERE.

Résumé / Summary

Functional brain data are often represented as a network or graph to model the brain regions as nodes and the connections as edges. A graph is an abstract object to represent multivariate data on a simple map of connections. In this tutorial, I will give a short introduction on the modelling of data as a network, with references. In a first part, I will describe precisely how to construct the connectivity networks using brain data recordings. Robustness and reproducibility will be discussed precisely. In a second part, I will describe tools to compare and classify the networks based on statistical tests or learning methods. Finally, I will conclude with a practical example from the clinical data on coma patients.
Related papers:
  • S. Achard, R. Salvador, B. Whitcher, J. Suckling, E. Bullmore. A Resilient, Low-Frequency, Small-World Human Brain Functional Network with Highly Connected Association Cortical Hubs. In Journal of Neuroscience, 26(1) pages 63-72, Jan. 2006.
  • S. Achard, C. Delon-Martin, P. E. Vértes, F. Renard, M. Schenck, F. Schneider, C. Heinrich, S. Kremer, and Edward T. Bullmore. Hubs of brain functional networks are radically reorganized in comatose patients. Proc Natl Acad Sci U.S.A., 109(50):20608-13, 2012.

Programme / Program

Voici le programme prévisionnel sur 2 jours / Here is the 2-day schedule:

3 Apr. 2017
09h30 Accueil / Welcome
10h-12h Cours sur les graphes (définition, exploitation de graphes, à quoi ça sert, comment les comparer et faire de la classification) - Discussion et questions
Lecture on graphs (definition, running, use, how to compare and classify them) - Discussion and questions
12h-13h30 Déjeuner / Lunch
13h30-15h30 TP sur le logiciel R. Travail avec un jeu de données réelles fourni pour la formation (1 groupe de contrôles, 1 groupe de patients) - But: mettre en application les outils vus le matin
Practical session with R software. Work with a real data set provided for the training course (1 control group, 1 patient group) - Objective: implement the tools seen in the morning
15h30-15h45 Pause/Break
15h45-16h45 Cours sur les statistiques sur les graphes
Course on graph statistics

4 Apr. 2017
9h-11h Cours sur la construction des graphes à partir de séries temporelles (outils mathématiques: ondelettes, corrélation, tests multiples)
Lecture on graph construction from temporal data (mathematical tools: wavelets, correlation, tests)
11h00-11h15 Pause/Break
11h15-12h15 TP avec R
Practical session with R
12h15-13h45 Déjeuner / Lunch
13h45-14h45 Cours pour lier les deux journées (point sur le seuil choisi pour les matrices de corrélation, enjeu et difficultés, autres outils)
Lecture to make the link between the two days (Point on the threshold chosen for the correlation matrices, stake and difficulties, other tools)
14h45-15h00 Pause / Break
15h00-17h00 TP avec R (faire toute la chaine de la construction du graphe jusqu'à l'exploitation statistiques des graphes) - Travail à nouveau avec les mêmes jeux de données que la veille, mais cette fois, les séries temporelles seront mises à disposition.
Practical session with R (Make the whole chain of the construction of the graph until the exploitation graph statistics) - Work again with the same datasets as the day before, but this time the time series will be made available.

Formatrice / Trainer

Sophie Achard, GIPSA Grenoble.

Organisateurs / Organizers