Dear Carl Maybe you can try the group (normal vs patient) * covariate (the volume of your ROI) interaction model in SPM. You could test the correlation pattern between two study groups. The following reference may be useful for your research. http://www.sciencedirect.com/science/article/pii/S1053811909000950http://www.ncbi.nlm.nih.gov/pubmed/18554926http://www.jneurosci.org/content/25/36/8303.full Hope it helps Best Paul > Date: Fri, 23 Nov 2012 14:44:06 +0000 > From: [log in to unmask] > Subject: [SPM] Between-group difference in structural connectivity (VBM) > To: [log in to unmask] > > Dear SPM/VBM users, > > I have performed a VBM T-test identifying an area of grey matter volume difference between two groups. Then I made an analysis of 'structural connectivity' by taking the voxel value at the peak voxel coordinate for each subject and using it as a regressor in a linear regression model for each group. Here I found another area that was positively correlated to the first one. The finding was present in both groups but the cluster differed in size and p-value. > > How can i compute the difference in correlation of the two areas between the two groups? Subtraction of betas or thresholded T-maps ? A completely different model? > > I am grateful for any suggestions! > > / Carl Johan Ekman > MD, PhD-student > Karolinska Institutet > Stockholm, Sweden