> I was wondering if there is a straightforward way to change from the
> default 12-parameter affine normalisation in SPM2 to test the effect of
> using a 9-parameter affine normalisation.
No. I'm afraid not.
>
> I'm working with T1 weighted, lesioned MR scans and often after affine only
> spatial normalisation, the original unnormalised images are in much closer
> resemblance to the template than the normalised images (with and without
> lesion masking).
Make sure that the contrast in your images is similar to that of the templates.
If the relative intensities of the different tissue types differs, then the
spatial normalisation will be biased.
Note also that the registration is regularised, so that the deviation of the
zooms and shears (away from those typically needed to register an average brain
to an MNI template) are penalised. You could try increasing this penalty
by changing spm_affreg.m from around line 392. The default regularisation
uses:
case 'mni', % For registering with MNI templates...
mu = [0.0667 0.0039 0.0008 0.0333 0.0071 0.1071]';
isig = 1e4 * [
0.0902 -0.0345 -0.0106 -0.0025 -0.0005 -0.0163
-0.0345 0.7901 0.3883 0.0041 -0.0103 -0.0116
-0.0106 0.3883 2.2599 0.0113 0.0396 -0.0060
-0.0025 0.0041 0.0113 0.0925 0.0471 -0.0440
-0.0005 -0.0103 0.0396 0.0471 0.2964 -0.0062
-0.0163 -0.0116 -0.0060 -0.0440 -0.0062 0.1144];
Increasing certain elements on the diagonal of isig could be used to
penalise shears rather more.
Best regards,
-John
--
Dr John Ashburner.
Functional Imaging Lab., 12 Queen Square, London WC1N 3BG, UK.
tel: +44 (0)20 78337491 or +44 (0)20 78373611 x4381
fax: +44 (0)20 78131420 http://www.fil.ion.ucl.ac.uk/~john
|