Multivariate approach for brain decomposable connectivity networks

Florent Chatelain (1) , Sophie Achard (2) , Olivier J.J. Michel (2) , Cedric Gouy-Pailler (3)

1 GIPSA-SAIGA - GIPSA - Signal et Automatique pour la surveillance, le diagnostic et la biomécanique
2 GIPSA-CICS - GIPSA - Communication Information and Complex Systems
3 LIST (CEA) - Laboratoire d'Intégration des Systèmes et des Technologies

This paper deals with the analysis of brain functional network using fMRI data. It recapitulates the concept of decomposable connectivity graph. Graphs are a usual tool to represent complex systems behavior, although edge strength estimation issues have not yet received a universally adopted solution. In the framework of linear Gaussian instantaneous exchanges, the well known partial correlation is usually introduced. However its estimation remains a challenge for highly connected or dense systems. Here, we propose to combine a wavelet decomposition and a graphical Gaussian model approach relying on decomposable graphs. This is shown to improve the estimations of brain function networks in the presence of long range dependence; the results are compared to those obtained with classical partial correlation estimators.

Mots clés: functional MRI; brain connectivity; partial correlation; graphical Gaussian model; maximum likelihood; decomposable graphs

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Florent Chatelain, Sophie Achard, Olivier J.J. Michel, Cedric Gouy-Pailler. Multivariate approach for brain decomposable connectivity networks. SSP 2011 - 2011 IEEE Workshop on Statistical Signal Processing, Jun 2011, Nice, France. pp.817-820. ⟨hal-00639853⟩