Hi,

I've just started using Freesurfer, and after running pre-processing steps with recon-all, some of my images have errors recognizing pial and WM surfaces.
Specifically, the most frequent problem is that pial and WM surfaces enclose part of the skull (which has not been removed very well).

Please find attached two images with examples of what I say. If you cannot see them as an attachment, they're also available at:

http://s5.postimage.org/os3auap5z/C801.png
http://s5.postimage.org/5oekrp1pj/A202.png

It seems like a skull stripping problem; some commands I've implemented to try to fix it are based on re-introducing the coordinates of brain centers by hand:

mri_watershed -atlas -c [i j k] T1.mgz brainmask.auto.mgz
cp brainmask.auto.mgz brainmask.mgz
recon-all -autorecon2 -autorecon3 -subjid [subject_name]

or:

mri_watershed -atlas -c [i j k] nu.mgz brainmask.auto.mgz
mri_normalize -g 1 brainmask.auto.mgz T1.mgz
cp T1.mgz brainmask.mgz
recon-all -autorecon2 -autorecon3 -subjid [subject_name]

(They're, of course, addressed to the right directories.)

The first set of commands has worked fine for a minority of all my images.
Nonetheless, there are plenty of subjects for whom neither of these scripts have gone well.

Should I solve it by manually adjusting pial/WM surfaces?
Or... do you have any idea on how to automatically solve it? (New scripts are welcome! :D )

Thanks in advance,
Aldo