[log in to unmask] wrote:
>
> Dear SPMers,
>
> we are working on SPECT images. We have 10 subjects analyzed in 2
> different conditions: (a) and (b). The first problem is the realignment
> step. This is what we did:
>
> (1) Number of subject: 10 -----> we selected individual scans
> (2) Which option? -----> Coregister & Reslice
> (3) Reslice inter.... -----> Sinc Interpolation
> (4) Create what? -----> All images + Mean images
>
> Results of this step are r*.img(hdr) and mean*.img(hdr) for each subject.
> What is the meaning of individual mean images?
see commnets in spm_realign_ui.m, page 2:
% Note that for PET, the coregistration is a two step process.
% First of all, the images are all realigned to the first in
% the series. A mean of these realigned images is created, and
% a second pass realignment is performed to realign all the
% images to the mean. Finally, the parameters are corrected
% for any differences estimated by registering the first image in
% the series to the mean image.
>
> The second problem is the coregister step. This is what we did:
>
> (1) Number of subjects: 10
> (2) Which option? -----> Coregister & Reslice
> (3) Modality of first target image? -----> target- SPECT
> (4) Modality of first object image? -----> object- SPECT
>
> Then we have:
>
> select target image for subject 1 ---> template SPECT.img
> select object image for subject 1 ---> meanimage for subject1
> select other images for subject 1 ---> *.img for all subjects
^^^^^^
Do you certainly mean "*.img for all subjects"? Don't you mean
"all scans for 1st subject"?
You can't apply transformations counted for 1st subject to another
subject, because they are not realigned one to another.
>
> The same for subject n. 2...10.
>
> The coregistration was wrong. Where is the mistake? In the realign step?
> In the coregister step? Or in both?
>
> Best regards,
> Michela Canfora
best regards
Tomas Sieger
|