Mutual information-based feature selection enhances fMRI brain activity classification

Vincent Michel (1, 2) , Cécilia Damon (1, 2) , Bertrand Thirion (1, 2)

1 PARIETAL - Modelling brain structure, function and variability based on high-field MRI data
2 LNAO - Laboratoire de Neuroimagerie Assistée par Ordinateur

In this paper, we adress the question of decoding cognitive information from functional Magnetic Resonance (MR) images using classification techniques. The main bottleneck for accurate prediction is the selection of informative features (voxels). We develop a multivariate approach based on a mutual information criterion, estimated by nearest neighbors. This method can handle a large number of dimensions and is able to detect the non-linear correlations between the features and the label. We show that, by using MI-based feature selection, we can achieve better perfomance together with sparse feature selection, and thus a better understanding of information coding within the brain than the reference method which is a mass univariate selection (ANOVA).

Mots clés: Features selection; Mutual information; Brain reading; Classification; fMRI

Domaines: Imagerie médicale; Informatique; Apprentissage [cs.LG]

Vincent Michel, Cécilia Damon, Bertrand Thirion. Mutual information-based feature selection enhances fMRI brain activity classification. 2008 5th IEEE international symposium on biomedical imaging: From nano to macro, May 2008, Paris, France. pp.592 - 595, ⟨10.1109/ISBI.2008.4541065⟩. ⟨hal-00504092⟩