Hey Naroa,
> We are trying to perform a longitudinal study with VBM and
> TBSS. We already processed all the images, but now we're not
> sure on how designing the matrix and the contrast to run the
> randomise. We have 2 groups (patients and controls) and 2
> times (base-line and follow-up). We're are interested in
> looking at changes in each of the groups over thime and to
> the interaction between time and group. How should the
> matrix and the contrast look like?
We had a discussion with Tom recently and I think the easiest way to do this in randomise at the moment would be to take *for each subject* the difference between the two timepoints (subject 1: timepoint1 - timepoint2; subject 2: timepoint 1 - timepoint 2; etc.) and then do something like (if you have, say, 2 controls and 3 patients):
Matrix:
1 0
1 0
0 1
0 1
0 1
Contrast:
1 0 (decrease over time for the controls)
-1 0 (increase over time for the controls)
0 1 (decrease over time for the patients)
0 -1 (increase over time for the patients)
1 -1 interaction between timepoint and group
-1 1 the other direction for the interaction
Hope this helps,
Gwenaelle
--------------------------------------------------------------------
Gwenaëlle Douaud, PhD
FMRIB Centre, University of Oxford
John Radcliffe Hospital, Headington OX3 9DU Oxford UK
Tel: +44 (0) 1865 222 523 Fax: +44 (0) 1865 222 717
www.fmrib.ox.ac.uk/~douaud
--------------------------------------------------------------------
|