Skull stripping and other EPI & MPRAGE registration issues

Dear AFNI expert,

We recently used 7 T MRI focusing on the basal ganglia.
Because I never had experience in preprocessing 7 T images, I have a lot of questions for preprocessing.

First of all, the attached image (OriginalvsUnifized) shows how T1 has been changed after using 3dUnifize.
After this, I would like to remove the skull, but the algorithm does not work well due to background noise.
Normally, in the 3 T image, there was no signal in the background (black), but here as you can see there are a lot of noises in the background.
I think this is why a vanilla version of 3dSkullStrip does not work.

Second, we achieved 1.1 mm isotropic resolution while focusing on the basal ganglia with narrow FOV.
For the registration between EPI and MPRAGE, should we need to use the same slab for the MPRAGE?

Sungshin Kim

Hi, Sungshin-

I would give @SSwarper a try on your original dataset.

It should produce a skullstripped+unifized version in the original space; a nonlinear transformation to standard space (e.g., MNI), and a representation of the brain warped to standard space. It can be a bit slow-- a couple hours on a laptop-- but if you can run on a multicore machine, you can specify the number fo threads to use (it is parallelized using OpenMP) by setting the environment variable OMP_NUM_THREADS.

Assuming you have an uptodate version of AFNI (“afni -ver” to check, “@update.afni.binaries -d” to update), you can run something like:

@SSwarper          \
        -input  DSET_T1_ORIG        \
        -base   MNI152_2009_template_SSW.nii.gz        \
        -subid  SUBJ_NAME        \
       -odir   .

… and please check the help file for more info (e.g., file names, automatic QC images made, etc.).