Hi Will,
The template is probably of a higher resolution than your fmri data, which means that registering to it will produce a larger file due to the increased number of voxels. This is one of the reasons that FEAT does all first-level processing on the native data and then registers only the resulting stats images, which are 3D and hence much smaller.
Most of the FSL tools don’t assume that functional data have been registered to a template. If you need to do it for your own analysis, there isn’t that much than can be done, but you might want to try running bet first to at least remove the non-brain voxels. That usually makes the files quite a bit smaller.
Hope that helps,
Eelke
> On 3 Aug 2016, at 16:32, William Cottam <[log in to unmask]> wrote:
>
> Hi,
>
> I'm running flirt from the command line calling previously defined registration .mat files. However, the output file is very large (over 200 mb) per subject when the original file is only 40 mb. Can someone please explain why this is happening and how to correct it (if possible). This is registering fmri (4D) data to MNI space.
>
> flirt -in input.nii.gz -ref subject.feat/reg/standard.nii.gz -out subjectMNI -init subject.feat/reg/example_func2standard.mat -applyxfm
>
> Cheers,
> Will
|