Hi, the "group" designator is for FEAT only, in order to tell FEAT that two separate variances might be present. Randomise does not care for these things. In order to account for repeated-measures, however, you need to manually edit the dual_regression script where randomise is called (the last step of DR). See here for details: http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Randomise#Accounting_for_Repeated_Measures You could also run randomise manually on the output of stage 2 of dual regression to speed things up a little. Hth Cornelius On Fri, Feb 22, 2013 at 1:40 PM, Carole Guedj <[log in to unmask]> wrote: > Dear FSL users, > > I'm a new user to FSL, and I'm a bit lost... In fact I'm not certain of my > analysis, and I'd like to be sure I process my data well. > I have two groups: sessions pre-traitement and sessions post-traitement (8 > vs 12 sessions). > > 1)I ran an ICA on all my data using -concat option: > *melodic -i session1.pre.nii,session2.pre.nii....,session1.post.nii.... > -a concat -o ../Output_ICA/all.group -m mask.nii -v --report --tr=2.000 > --Oall * > > 2)I ran a dual_regression: > *dual_regression group/melodic_IC.nii.gz 1 glm.mat glm.con 500 output > NIFTI_files/session.** > * > * > the matrix I used: > *goup EV1 EV2* > * 1 1 0* > * 1** 1** 0* > * 1** 1** 0* > * 1** 1** 0* > * 1** 1** 0* > * 1** 1** 0* > * 1** 1** 0* > * 1** 1** 0* > * 2** 0** 1* > * 2** 0** 1* > * 2** 0** 1* > * 2** 0** 1* > * 2** 0** 1* > * 2** 0** 1* > * 2** 0** 1* > * 2** 0** 1* > * 2 0** 1** * > * 2** 0** 1* > * 2** 0** 1* > * 2** 0** 1* > * > * > * > * > the contrast I used: > *B-A 1 -1* > *A-B -1 1* > *mean A 1 0* > *mean B 0 1* > * > * > -Is that correct? > -How the program knows how to separate the two groups? I mean, how the > program knows that this session is a pre-treatment session and another one > is a post-treatement session? Is it the matrix that allows this? > -And to compare the two groups, is that I have to run another analysis or > I use only the outputs : dr_stage3_ic00##_tfce_corrp_tstat1.nii and > dr_stage3_ic00##_tfce_corrp_tstat2.nii > > Hopefully someone can help me and thanks in forward... > > > Carole > * > * > * > * > * > * > * > * > * > * > -- Cornelius Werner cornelius.werner<at>gmail.com