Hi,
You can either just use the T1_orig as the reference volume in step 2 (as the reference volume controls the resolution and FOV of the output, and is used for nothing beyond this when applying transformations) or you could use the ApplyXFM GUI instead (for step 2) as this GUI allows you to explicitly set your resolution and FOV.
You should not need to use -usesqform when applying transformations, and I think this might be what is causing the problem.
All the best,
Mark
On 2 Sep 2014, at 00:38, Christopher Coello <[log in to unmask]> wrote:
> Hi all,
>
> I know this is a well-described subject in this mailing list, but I don't find a clear and direct command line answer to the following case.
> I want to :
> 1/ calculate the transformation matrix using FLIRT an individual T1 MRI (T1_orig) to the MNI 1mm (MNI_1mm)
> 2/ apply this transf. matrix to T1_orig and other images in the same space as T1_orig without resampling to the isotropic 1mm3 voxel of the MNI_1mm volume.
> For now these are my comand lines :
> for 1/ -> flirt -in T1_orig -ref MNI_1mm -omat trMat (with or without -usesqform)
> for 2/ ->
> flirt -in T1_orig -ref MNI_1mm -out T1_reg_1mm -init trMat -applyxfm (with or without -usesqform)
> followed by
> flirt -in T1_reg_1mm -ref T1_orig -out T1_reg_origmm -applyxfm -usesqform
>
> The result of such procedure will produce a register volume in the appropriate space (volume dimensions and pixel size) but it seems with a wrong origin, the last step making the origin to be reset to the T1_orig origin.
> I have tried to use fslcpgeom T1_reg_1mm T1_reg_origmm -d without any significant change in the result.
> I also tried with and without -usesqform as not really understanding the purpose of this option in the first two commands, but without success either.
>
> Any idea of further tests ?
>
> Thanks in advance
>
> Christopher
|