Dear experts, I have run a VBM analyses for three age groups. Now, I want to correlate its fstat output with a cognitive variable (for example, performance in a short term memory task). I have thought the following: 1) Calculate a binary mask for the fstat output with: fslmaths <fstat results> -thr 0.999 -bin <mask name> 2) randomise with the correlation design using the fstat mask instead of the GM_mask randomise -i GM_mod_merg_s3 -m fstat_mask -o <name> -d correlation.mat -t design.con -T -V In this way I expect to calculate the relationship between those GM regions where the age had a significant effect and the performance in my cognitive variable. Is it correct? Then, I want to go on with the post hoc analyses output. Specifically, I do not want to go with the fstat output now but with the tstat3 output. So, now, I will not include the whole sample, but only two groups. So, as I guess I need to calculate a new mask with the tstat3 output and then run a new randomise for the correlation. But, I guess I should select only the subjects of these two groups from the GM_mod_merg_s3. Is it correct? How could I do that? thank you very much in advance and a happy 2012 Daniel Ferreira