Hi Darren,
I think if you have a q/s-form matrix with off-diagonal elements, then
your images must actually be SPM5 NIfTI images (although possibly with
.hdr/.img pairs rather than .nii single-files), rather than old SPM<=2
Analyze.
I wonder if you have your FSLOUTPUTTYPE set to Analyze? I think that
would be forced to drop the rotations etc from the q/s-form, since
Analyze can't store them. Try setting it to NIFTI_PAIR or NIFTI.
Hope that helps,
Ged
On 26/10/2007, Darren G <[log in to unmask]> wrote:
> Dear FSL'ers
>
> I have been trying to use BET2 on analyze files made in SPM. I must be doing
> something worng as the problem is that after extracting the brain the
> orientations are wrong (i.e., different) on the extracted vs. original images.
>
> Here is the output of fslorient for an SPM-ANALYZE file
>
> fslorient -getsform s1094-0004-00001-000001
> 0.85879 0.0316479 -0.00212486 -82.5151 -0.0314743 0.85715 0.0618951 -125.362
> 0.00378017 -0.053088 0.99808 -60.3387 0 0 0 1
>
> The exact same numbers are returned for fslorient -getqform
>
> Now I run bet2.
> bet2 s1094-0004-00001-000001 s1094-0004-00001-000001_brain
>
> Although fslview can display the original and extracted images
> (s1094-0004-00001-000001 and s1094-0004-00001-000001_brain) as aligned, SPM
> does not. So I rechecked the orientations:
>
> fslorient -getsform s1094-0004-00001-000001_brain
> -0.859375 0 0 77.3438 0 0.859375 0 -123.75 0 0 1 -67 0 0 0 1
>
> fslorient -getqform s1094-0004-00001-000001_brain
> 0.859375 0 0 0 0 0.859375 0 0 0 0 1 0 0 0 0 1
>
> So the orientation appears to have been changed by running the image through
> bet extraction.
>
> I am running FSL within the VMware player.
>
> Please let me know what I am doing wrong and how to avoid this.
>
> thanks,
> darren
>
|