Dear Thomas,
> I have a few questions regarding group analysis with the Random
> Effects Model in fMRI.
> We perform an experiment with a block design as follows:
> 2 sessions per subject, 3 stimulation conditions (Task A, Task B, Rest R),
> session 1: R A R A R A ...
> session 2: R B R B R B ...
>
> We are interested in the differences of the activations from the two tasks,
> as well as in what they have in common.
> Therefore we would like to look at:
> - both conditions against rest (which is not a problem)
> - the Conjunction analysis of A and B (which was possible using the fixed
> effects model)
Why is this not possible at a first-level analysis? The R scans are
different for the two contrasts entering inot the conjunction and
therefore these contrasts are orthogonal.
> How can we do a conjunction analysis of A and B using SPM99 and the Random
> Effects Model ?
If you assume sphericity then you could enter the A-R and B-R contrasts
images into a second level and test for the conjunction of [1 0] and [0
1]. However, because SPM always models a constant term you are obliged
to enter a covariate of [1 1 ...1, -1 -1 ... -1] and test for a
conjunction of contrasts [1 1] and [-1 1].
> How can we do the subtraction A - B when A and B are acquired in
> different sessions from one subject ? I heard that there could be a
> problem with the global scaling if A and B are not acquired in the same
> session.
At the second level this should not be a problem (the data are globally
normalised at the first level). To compare A and B, relative to rest,
simply perform a two sample t test on the contrasts [1 -1] and [1 -1]
from the first level (i.e. A-R - B-R = A-B).
I hope this helps - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|