 |
 |
You can reslice the template (with the adjusted matrix) so that it matches one of your normalised images. The reslice only option of coreg is good for this. Not sure this is the easiest way to find outliers though. I'd try a PCA to see if there's anything to be seen there. Within the Dartel toolbox, there's an option to "build kernel from images" (or similar). This will save a file called dp_*.mat. If you load this file, you'll see it contains a square matrix that is the same size as the number of scans you specified. You can run svd on this matrix, and plot the first few columns of the results. [U,S]=svd(Phi); plot(U(:,1:4)) If there's nothing obvious in there, then try: plot(U(:,5:8)) etc Best regards, -John On 15 July 2011 10:40, Katharina Wittfeld <[log in to unmask]> wrote: > Thank you very much, John! > If I now check with "check reg" my Template is scaled like the one coming with the VBM8-Toolbox. > > But I've some more questions. Sorry for asking such basic things. > I've followed your VBM Tutorial from 2010 to apply DARTEL to my data set (only structual T1 data). As Output I've the smoothed Dartel normalized swc1*.nii images which are in MNI152 and I've my Template_6. > If I want to compare the volumes of the subjects with the Template to find outliers. Let's say for example pairewise by some kind of correlation. I need the volumes to be comparable. Just replacing the matrix doesn't help in this case. > So I'm looking for a way to apply this affine transform. > > But maybe there are much better ways to find outliers in an automated way. Any hint is very much appreciated since the data sets are quite big. > > Katharina > > >