If you used resliced anatomical images to do the coregistration, then this
would have been a bad idea. Just use the original anatomical images to
estimate the transformation (because the matrix in the header/.mat file)
should take care of the original positioning.
I'm not sure which version of SPM you used, but SPM5 segmentation should give
more accurate spatial normalisation parameters.
Best regards,
-John
On Saturday 15 July 2006 23:02, Marina Papoutsi wrote:
> Dear all,
>
> I was hoping that someone can point me to some solution with a problem I
> am having with normalization.
>
> I have acquired oblique, partial brain functionals and anatomicals in
> the same position like the functionals. The functionals have a 224 fov
> (2.3x2.3x2.3mm), while the anatomicals 220 (0.859x0.859x2.3). I am able
> to properly coregister the functionals to the anatomicals and I then use
> the anatomical to estimate the parameters for normalization. After
> normalization however, the output images (w*) have been changed and have
> become much bigger. The original images started at about the level of
> the ac and the normalized images cover the whole prefrontal cortex and
> the superior portion of the cerebellum and appear deformed somewhat at
> the inferior-posterior portion.
>
> My question, is there any way I can achieve better normalization than
> that? I have tried different interpolation degrees and different
> estimation approaches and I have even created a weighted template image
> to account for the fact that I am not imaging the whole brain (0 for the
> voxels not included). The results are the same. Is there any way I can
> achieve better normalization than this?
>
> Any help will be very much appreciated.
> Thanks,
> Marina
|