 |
 |
Did you install any updates to SPM5 between doing the two sets of processing? The discrepancy comes from the way the coordinates of the voxels in the spatially normalised data are selected to best fit the bounding box. I can't remember exactly what changes were made, but you can obtain the same image dimensions by setting the bounding box to: [[-78 -112 -50];[78 76 84]] instead of [[-78 -112 -50];[78 76 85]] . The difference depends on whether 85/2 (ie 42.5) should be treated as being closer to 42 or 43. Best regards, -John On 1 September 2011 14:48, André Schmidt <[log in to unmask]> wrote: > Dear SPM experts > > > > We have already normalised a set of PET images with SPM5 using a template > having the dimensions 79 x 95 x 68. Now we wanted to add a number of new > subjects. We used the same template to normalise the new data, but this > yielded normalised images with somewhat changed dimensions 79 x 95 x 69, so > that we could not pool the two samples for further analyses. In fact we got > a error message. We have no idea why this happens and how we could change > the dimension of the images. > > > > Hope someone could help us understanding where is the problem here. > > Thanks in advance > > > > André > > > > > > > >