The first step in the segmentation process is an affine registration to MNI space. In SPM2, this was done by minimising the mean-squared difference between the image and a template of the same contrast. In SPM5, a mutual information registration to tissue probability maps is used. As there are only tissue probability maps of GM, WM and CSF and nothing for skull, scalp etc, then this is not always stable. Because of this, it can be a good idea to have your origins set reasonably close to the AC, and the orientation of the images close to that of the template data. Best regards, -John On Monday 22 October 2007 15:14, Susan> <Shear wrote: > We are able to segment T1 images including post-contrast images using > SPM2, but segmentation is corrupted or fails in SPM5 with the post- > contrast images. Does any one have any suggestions on how to fix this > problem?