Dear Lucas, Yes, you can use ImCalc for this. Just make sure to change the data type of the output image to float32. Not sure I understand your alternative. You can implement ImCalc with a GLM - create a design matrix being the identity matrix with the input images being subj1-pre subj1-post subj2-pre subj2-post ... and use contrasts [1 -1 0 ...], [0 0 1 -1 0 ...] to create the summary statistics images (identical to ImCalc above) entering the two-sample t-test. Best regards, Guillaume. On 23/01/17 15:38, Lucas L. wrote: > Dear Guillaume > > Thank you again for your both answers. Regarding the first one (below), just to be sure I understand you correctly: > > may I use ImCalc (i1-i2) to compute the differences in pre-post intervention at the single subject level and then enter these images in a two-sample t-test to compare the main effect of the intervention between both groups? Or would it be equally valid to first run a paired t test for the pre-post intervention for each group separately and then compare the difference between groups using ImCalc with paired t test tmaps as input? > > Kind regards, > Lucas > > > >>> Compute the main effect of your within-subject factor (intervention, ie > pre-post) for each subject and enter these images in a two-sample t-test > (for the between subject factor, group). A contrast [1 -1] will test for > a group by intervention interaction. >>> > -- Guillaume Flandin, PhD Wellcome Trust Centre for Neuroimaging University College London 12 Queen Square London WC1N 3BG