| Paper: | SA-PM-PS3.5 |
| Session: | Functional, Dynamic and Parametric Imaging |
| Time: | Saturday, April 8, 13:30 - 14:50 |
| Presentation: |
Poster
|
| Title: |
Robust Statistics for Nonparametric Group Analysis in fMRI |
| Authors: |
Sébastien Mériaux; CEA, Service Hospitalier Frédéric Joliot | | |
| | Alexis Roche; CEA, Service Hospitalier Frédéric Joliot | | |
| | Bertrand Thirion; INRIA Futurs | | |
| | Ghislaine Dehaene-Lambertz; INSERM U562, Service Hospitalier Frédéric Joliot | | |
| Abstract: |
In order to deal with inhomogeneous populations of subjects in fMRI studies, we propose several robust statistics to perform group analysis: sign statistic, Wilcoxon's signed rank statistic and empirical t-statistic. The test's significance levels are calibrated using the same sign permutation framework as in Holmes et al. (1996), allowing for exact specificity control under a mild symmetry assumption about the subjects' distribution. The benefit of the robust approach is illustrated with group analysis results obtained from a speech listening experiment involving three month-old infants. |