Hi Jan, It seems that you actually would like to see whether there is a change between post and pre-intervention scans that differs between groups. It isn't just if post-intervention scans differ between groups, but if when compared to the baseline (pre-intervention) the groups differ. You can use the same design as in the previous email, with the subtractions in fslmaths, and if there isn't a score, or any other variable that could be a nuisance (e.g., age, sex, disease severity, etc), you can drop the EV4, and modify the contrast accordingly. Regarding the what goes in randomise, don't use the *_Z.nii.gz images, because these are z-scores, which don't possess the same variability of the original data, which you need to accurately make inferences on that data. Use instead the non-Z versions of the same files. Hope this helps! All the best, Anderson On 25 September 2014 21:52, Jan B <[log in to unmask]> wrote: > Hi Anderson, > > thank you very much for your quick response and the valuable suggestions! > What I really want to know is whether the post-intervention means differ > between groups when adjusted for the pre-intervention means (my question > was unclear, sorry for that). So the solution is - as you decribe in the > 2nd answer - to run an ANOVA using the within-subject differences > calculated with fslmaths as inputs. > Do you think that it is appropriate to use the individual > dr_stage_2..Z.nii.gz files for the difference calculation (they have been > all generated by one dual regression analysis)? > > All the best, > Jan > >