 |
 |
Dear FSL users, Has anyone any suggestion regarding my last Email? Many thanks in advance, Best, Stephane Le 25.07.13 14:03, Stephane Jacobs a écrit : > Hello, > > I am running an ROI analysis on data analyzed with FEAT following a 3 level procedure (run, subject, group). I wanted to define my ROIS based on the clusters obtained at the group level, and then find, at the second level, the individual peak of activity within that ROI for each subject independently to define smaller ROIs that would take into account inter-individual anatomical variability, and then extract copes from independent functional runs in those regions at the first level. > However, some of my subjects don't show a significant effect for the contrast used to define the ROIs in one or several selected regions. I was thus wondering what would be the statistically soundest way to deal with these subjects: > > 1. exclude them from the ROI analysis for the regions that are not activated in their 2nd level analysis. > 2. use the group peak when no individual peak is detected. > 3. define larger ROIs at the group level, and extract the copes from the exact same mask for all subjects. > > Also, is there any recommendation as to what a reasonable size would be for an individual- or group-based ROI? > > Many thanks in advance for you help! > > Best, > > Stephane >