Hi James
> Hi,
>
> I am relatively new to SPM, and have been having problems normalizing an
> anatomical and corresponding functional images. I'd really appreciate any
> advice you have to offer.
>
I'll try :-)
> 1. I registered my functionals to motion correct.
>
ok - no slice timing?
> 2. I corregistered the anatomical to one of my functionals
>
I would corregister with the mean functional (when doing the realignment
only estimate the parameters and write the mean functional)
> 3. I attempted to segment the co-registered anat to get the tissue
> probability maps and the *seg_sn.mat. The grey and white matter images came
> out very distorted.
> 4. I attempted to normalize:write my registered functionals using the
> *seg_sn.mat parameters, and this produced terribly distorted images.
>
no use ; when normalizing SPM takes care of everything
simply use normalize and select the coresgistered t1, the t1 template
and apply to the functional, that's it
> 5. I checked the display corregistration between my functionals, anatomical,
> and the EPI, T1, and tpm templates, and found that there was a large
> translation discrepancy between all of my images and the template images.
> 6. I proceeded to reorient my anatomical (via translation) using the
> reorient tool on the display window, and saved the reorientation.
> 7. I tried to do segmentation on my anatomical image again, and this worked
> well.
> 8. I attempted to normalize my functionals using the new *seg_sn.mat. This
> time, my normalized functionals were not distorted (a big improvement!),
> however these functionals were cut-off on one side. When I displayed the
> images, they took up the entire display space. I tried to translate the
> image to the side and bring the missing part of the brain into view, but
> found that the brain was simply cropped--the entire prefrontal area cut off
> from all my functionals.
>
note that cut off may appear on the bottom and top of the functionals
(typically half slice or one full slice); this is explained by the fact
that you want that all brain images have the same size, thus losing
sometimes some part (see realignment)
if the idea of the segmentation was for you latter one to work on the
grey matter only you can use a grey matter mask during the statistical
modelling, hence reducing the search space to grey matter.
Hope this helps
cyril
|