Dear SPMers,
I wonder how to perform the conjunction analysis at the second level in
spm2. Or conjunction analysis can not used at the group level with random
effect model.
If it can do, follows are some background and procedure. We appreciate
your comments.
We perform a fMRI experiment with a block design as follows: 2 sessions
per subject (totally 12 subjects), 3 stimulation conditions (Task A, Task
B, baseline b),
session 1: A b A b A b A b A b
session 2: B b B b B b B b B b
We conducted the design matrix at the first level with implicit
baseline, i.e., for A-b contrast, [1 0 0 0], for B-b contrast [0 1 0 0].
Then we put all the 24 con* images (12 for A-b,then 12 for B-b) into
a "One Way ANOVA" analysis (number of groups: 2)within "basic models" of
spm2 . Then define [1 0] for A contrast,and [0 1] for B contrast. We test
for a conjunction of the two contrasts (by "Control" selected), and
inclusively masking with the two contrasts with p<0.001. We get the final
result with threshold p<0.001 (uncorrected).
I wonder if the above procedure works to get the common activations by
A and B task, simultaneously eliminating the significant difference
between A and B.
Thanks a lot.
|