Dear Caroline,
I think the problem is that you are confusing a conjunction procedure with
inclusive masking. If you simply want to see the areas that were significant
in the three contrasts, use inclusive masking at the appropriate threshold.
A conjunction analysis (in SPM99 and SPM2) is a more sophisticated analysis
that tests for the joint significance of the three contrasts (any one of which
may not be significant in its own right). A conjunction analysis of this sort
is like an F-test; it furnishes an inference about multiple contrasts. The
special constraint in a conjunction analysis is that the responses all have to
be activations (i.e., positive responses).
These are a couple of recent papers in NeuroImage about this confusion you
may
find useful.
I hope this helps - Karl
At 09:14 17/05/2005 +0200, you wrote:
>Dear Dr Friston,
>
>Sorry for direct e-mailing you.
>But my question was not answered through the list.
>And I hope that you can help me out.
>
>Thanks in advance.
>caroline
>
>____________________________________________________
>
>
>Dear SPMers,
>
>I am confused with the function of the exclusive masking.
>
>I have 3 fmri sessions. Each has 3 conditions.
>Session A: Rest A1 A2
>Session B: Rest B1 B2
>Session C: Rest C1 C2
>
>They are put into the same design matrix as A1 A2 B1 B2 C1 C2 for 1st
>level analysis. The rest conditions for all 3 sessions are left out.
>
>Basically A1 and A2 are the same visual stimuli and the only
>difference is the stimulus for A1 is moving to the right and for A2 is
>moving to the left.
>It is the same for sessions B and C.
>A B and C are different visual stimuli.
>
>The contrasts are definded as follows:
>
>1 1 0 0 0 0 to look at A
>0 0 1 1 0 0 to look at B
>0 0 0 0 1 1 to look at C
>
>I would like to find the common and different areas that all 3
>sessions are activated.
>I use the aal toolbox.
>
>For the common areas:
>I use the function conjunction. And so far so good.
>
>BUT,....For the different areas:
>
>When I look at all sessions separately, ie comparing the activation
>areas manually, table by table, it is different from if I mask them
>with exclusively as A exl (B and C).
>
>eg. It reported the areas that are commonly activated in A B and C also.
>
>I am not cleared that should I use exclusive masking or should I just
>compare tables?
>
>Any comment would be appreciated.
>
>Thanks in advance.
>
>best regards
>caroline
>
>
|