Dear Mclaren,

Thanks but I'm not totally clear about your reply so I hope to get further confirmation and would appreciate your help. Take this case for example:

Example 1: A 2-sample t-test map to compare activation between groups A (10 subjects) and B (10 subjects).
Example 2: A conjunction analysis map of "activation in group A" and "activation A>B."
Example 3: A 1-sample t-test map of "activation in group A."

For Example 1, my procedures are:

(1). From "segment", I obtained wc1-A1, wc1-A2, ...wc1-A10, wc1-B1, wc1-B2, wc1-B10. In "Imcalc," I first averaged the 20 unmodulated normalized images using (i1+i2+....i20)/20 and then thresholded and binarized the average image using (i1>0.2). Then I got the binarized mean gray matter mask ave-gm.

(2). In "Result" then "whole brain", I noted down the smoothness of residual images of each subject's 1st-level fmri model. I then averaged the 20 data to get the average smoothness s1 s2 s3.

(3). I then used the command below to calculate cluster size threshold for this 2-sample t-test map.
     3dClustSim -mask ave-gm -fwhmxyz s1 s2 s3 -iter 10000 -pthr 0.001 0.005 0.01 -athr 0.05 -NN 2

Is any step incorrect?

And, for Examples 1 and 2, the cluster size threshold would be the same. For Example 3, it would be different from that in Examples 1 and 2. Is this correct?

Really thanks for help.

Mark