> We have an MR T1 study which has very different scalp/skull information > from the SPM'96 T1 template - i.e. there is scalp/skull missing in > region where it exists on the template and vice versa - will this induce > problems with a 12-point spatial normalisation? It would probably give you a spatial normalization that is not as good as it would be if the template looked more like your image. If you find the affine registration is not so good, it is possible that it is cought in a "local minimum". To fix this, I would suggest modifying the starting estimates (by the defaults button, or by specifying better values in the origin field of the headers) so that the program has a better chance of reaching a good solution. Regards, -John %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%