Dear FSL list,
The past days I have been struggling with my TBSS analyses, and feel quite desperate after trying numerous different things. Hopefully someone here can help me out:
I have one group of 49 participants, and would like to correlate their DTI data (FA maps) with a behavioral outcome.
In Glm I made a design matrix, with 49 inputs, and 2 EVs: the first is the group (only ones), which I understood has to be defined with a correlation, and the second EV containing the behavioral scores ('score').
Contrasts I defined:
1. Group mean [1 0]
2. Positive correlation score [0 1]
3. Negative correlation score [0 -1]
The first time I tried this, I had demeaned the behavioral outcome (which is a score between 0 and 1), and used the following command:
> randomise -i all_FA_skeletonised -o score_N49 -m mean_FA_skeleton_mask -d design_score.mat -t design_score.con -x --T2
It ran smoothly, but almost the whole skeleton appeared to be significant, even with TFCE, for contrast 2 (however no significance at all for contrast 3..). This seems very unlikely and unspecific, and I understood this could be an error. So from several sources I received the suggestion to add '-D' to the command.
I did this two times, one by demeaning the behavioral score, and once without (since I read in the mailinglist that this is also done during randomise, when giving the -D option), but I get the same error message both times:
> randomise options: -i all_FA_skeletonised -o score_N49 -m mean_FA_skeleton_mask -D -d design_score.mat -t design_score.con -x --T2
> Loading Data:
> Data loaded
> Warning: Data demeaning selected, but at least one design column has non-zero mean - therefore invoking automatic demeaning of design matrix
> One-sample design detected; sign-flipping instead of permuting.
> 5.6295e+14 sign-flips required for exhaustive test of t-test 1
> Doing 5000 random permutations
> Starting permutation 1 (Unpermuted data)
> Starting permutation 2
> Warning: tfce has detected a large number of integral steps. This operation may require a great deal of time to complete.
... an output that I have been seeing for over 2 days now, without anything changing..
I have been going through the randomise documentation and FSL mailinglist, but can't find a solution..
Anyone out there, who could help me out?
Thank you in advance!!
Best,
Zsuzsi
|