Hi,
What does "nonsense" mean?
Are the qform and sform *codes* set to non-zero values?
If not, it just does not matter what is in the matrices themselves.
I'm also not sure what you mean by "not a brain image".
Is all the problem caused because one method gives you a
result in standard space and the other gives you one in
native space? If so, you can just transform one of them
to be in the other space using the -applyxfm feature of flirt
and the matrix calculated by run_first_all.
I think I will need much more precise information about what you
are doing before I can help.
All the best,
Mark
On 6 Jan 2009, at 18:17, siamak wrote:
> Hi Mark,
>
> Thank you for your reply. I did what you said, but the results were
> nonsense.
> I checked the Q_form/S_form of all the files. In all the files,
> Q_form is the
> same as S_form. But, Q_form/S_form of each file is different from
> the other. I
> ran "fslcpgeom file1 file2", when I view file2, it is not a brain
> image, but some
> noise. I ran "fslcpgeom file1 file2 -d" , the result was brain image
> but is was
> streched or sheard, since file1 and file2 have different dimentions.
>
> Actually, I am using a registration framework before segmentation.
> Since my
> original image has not the same orientation and dimention as MNI
> reference
> template, the registered image has different dimention and orientation
> compared to my original image. If I segment this registered image by
> fsl, the
> segmented image indeed has different orientation and dimention
> compared to
> the original image. When I use fsl for both registration and
> segmentation using
> run_first_all command, I am not able to compare the results of the
> segmentation, since the first segmented image (my registration and fsl
> segmentation) has different dimention and orientation compared to
> the second
> segmented image (fsl registration and segmentation). May be this
> helps you to
> guide me more.
>
> Best,
> Siamak
>
|