Dear John,
yes, the instructions inside SPM are quite confusing... create 13 "subject" fields, for "Scans" select the corresponding individual five con images, and for "Conditions", enter 1 2 3 4 5. The vector for the F contrast should look like this
1 -1
1 0 -1
1 0 0 -1
1 0 0 0 -1
0 1 -1
0 1 0 -1
0 1 0 0 -1
0 0 1 -1
0 0 1 0 -1
0 0 0 1 -1
but it should be sufficient to go with
1 -1
1 0 -1
1 0 0 -1
1 0 0 0 -1
as the other rows are just combinations of the first five (second row minus first row is equal to the fifth row). This F contrast would pick up voxels that differ between any of the five conditions (which might be interesting or not depending on your research questions). To get p-values on cluster-level for F contrasts use SPM12b (think this hasn't been implemented into the final version of SPM8). Other comparisons between any of the conditions should be valid as well, so go with t-contrasts or F-contrasts like [1 -1] or [2 -1 -1].
Concerning "commonality", the only valid way for repeated measurements way is descriptive (null conjunctions are valid only for between-subject designs). Create simple contrasts like [1] for each of your your five conditions on single-subject level. [1], [0 1] was just an example if you had two different conditions A B within a single run within your single-subject model. Then run a first one-sample t-test across all your con_0001 images, a second across con_0002 images and so on. Within these models, create a contrast [1] (and if you're interested in, a second contrast [-1]) , then threshold it at a certain value. Using "save..." and "all clusters (binary)" you can export significant voxels and clusters as a nii file. Significant voxels will have a value of 1, the rest is 0. Now you can load the five nii files corresponding to your five conditions with MRIcroN, xjview or other viewing tools to look at the overlap.
You can also use SPM's "ImCalc" to create a single image. As input images, use the five nii files. As expression, use something like i1 + 2*i2 + 4*i3 + 8*i4 + 16*i5. Voxels which were significantly activated during all five conditions should have a value of 31. Voxels activated during conditions 1, 3 and 4 would have a value of 13 and so on. In a similar way you could add the five nii files for negative activations, but you have to make sure that any of the combinations of positive and negative activations has a unique value. With xjview you should be able to load this new image. Set the cluster size to 0 and intensity to for example 30.5. Using the "report" function you should be able to get the corresponding voxels / clusters, some MNI coordinates and anatomical labels.
Best,
Helmut
|