> I have some functionals converted from siemens .ima
> The information on origin is not avaliable for some reason, so the .hdr do
> not have any origin field. I tried to reorient the functional setting
> origin more or less at AC using EPI template as reference. My structurals
> are sagittal, so I reoriented them to be axials and set the origin to AC,
> and checked they were aligned to AC-PC line. Using check reg I can see
> T1.mni, my T1.img, EPI.mni, my functionals seem more or less aligned. Now,
> after slice timing, and realignment to 1 session I coregistered my
> structural with functional mean image using 1-target image
> :mean-functional-image
> 2-object image: T1 from same subject
>
> (does spm2 automatically set between modality coregistration?
The Coregister button uses normalised Mutual Information as an objective
function by default. This should work for both inter- and intra-modality
registration.
>
> After coregistration I check for alignment using check reg loading mean
> functional first in order to have that image to define space, and T1 from
> subject. I can see that origin is shifted a few mm toward superior and
> posterior with respect to AC.
It may move the origin slightly - but the images should (hopefully) be in
closer registration afterwards. The origin is usually not very clear on EPI
data, so it can be difficult to set precisely. Also, the presence of
nonlinear distortion and dropout in the EPI may be another factor to
consider.
>
> I go on with normalization using :
> 1- template: T1.mni
> 2-source: T1.img coregistered to functionals as above.
>
> Now If I use check reg utility to watch in following order T1.mni,
> EPI.mni, normalized wT1.img, normalized functionals, my functionals seem
> more or less aligned again.
>
> Why the coregistration step does change origin then?
Does the coregistration bring the functional and structural images into closer
alignment.
> I have some amount of distortion in my functionals, would unwarping help at
> coregistration step ?
If there is distortion in the functionals, then you should get a better
registration if this is corrected first.
Best regards,
-John
|