Hi all,
I think this is harder than it sounds, but perhaps I'm just confused...
I don't think imcalc is suitable, since expressions like .* assume
equally sized Matlab matrices, so I don't think it can be used for
resampling.
spm_reslice sounds like it is suitable, but from my reading of the
help, you need the correct voxel-world mappings in the .mats, which
will depend on the current voxel-world mapping as well as the desired
resampling factors... I've not really investigated spm_reslice though,
so I might be wrong about this, or missing something obvious.
For Harmut's approach
Hartmut Schütze wrote:
> ... if you just sample it down by applying some normalisation and set the
> defaults:
> defaults.normalise.write.vox = [1 1 1];
> defaults.normalise.write.bb = [[-100 -130 -80];[100 120 90]];
I think you can use these params with normalise write to get the
desired voxel sizes and bounding box, but you'll need an identity
sn.mat if you don't actually want to change the normalisation.
I've had a quick play normalising the T1 template to itself, the
attached sn.mat and norm_job.mat seemed to create a correct 1mm
isotropic version of T1.nii, but I haven't thoroughly investigated this.
Best,
Ged.
|