External Email - Use Caution
Hi Valeria,
you can account for the inhomogeneities Bruce was talking about by either using SPMs or FreeSurfers inhomogeneity correction. Personally, I prefer SPM over FreeSurfer, but if you supply a brainmask to the FreeSurfer tool, it gives reasonable results around the temporal lobes, too.
I haven't used an MEMPRAGE sequence, however, using an MPRAGE you can use a FWHM of 30 and a regularization of 0.001 in SPMs segmentation tool, setting the sample distance to 2 improves the correction further. Using the N3 implemented in FreeSurfer you could run (from the subject's directory) mri_nu_correct.mni --n 2 --uchar transforms/talairach.xfm --mask brainmask.mgz --stop 0.0001 --distance 15 --fwhm 0.15 --shrink 2 --proto-iters 10000 --i orig.mgz --o nu.mgz. Then proceed with the normalization of autorecon1. Both correction methods take quite some time.
In case of MP2RAGE or if the MPRAGE was divided by a GRE manually you will need to adjust these settings accordingly. The N4 algorithm of ANTs works fine also if a brainmask is supplied. In case someone wants to have some parameters to run an inhomogeneity with that tool or on MP2RAGE/Division data, let me know.
Best, Falk
-----Ursprüngliche Nachricht----- Von: freesurfer-bounces@nmr.mgh.harvard.edu freesurfer-bounces@nmr.mgh.harvard.edu Im Auftrag von Bruce Fischl Gesendet: Montag, 25. März 2019 20:27 An: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Cc: freesurfer-bounces@nmr.mgh.harvard.edu; Andre van der Kouwe andre@nmr.mgh.harvard.edu; Jonathan Polimeni jonp@nmr.mgh.harvard.edu Betreff: Re: [Freesurfer] recon from 7T data
Hi Valeria
Jon Polimeni and Andre van der Kouwe put together a specific acquisition and set of preprocessing steps for 7T (mp2rage I believe). You need to account for the transmit inhomogeneities, which are what is causing you to lose the temporal lobes.
cheers Bruce
On Mon, 25 Mar 2019, Barletta, Valeria wrote:
Dear Freesurfers,
I am trying to perform the FreeSurfer cortical reconstruction process from my 7 Tesla MEMPRAGE, but am getting very bad results (see first attachment) compared to what I had using the 3 Tesla images (second attachment).
Before launching the "recon-all ....... -all" command, I applied the FreeSurfer Gradient non-linearity correction.
What can have been the problem?
Thank you very much for your help,
Valeria