Kathy,

I received the example files that you uploaded. It looks like the problem you are encountering was one that was fixed in v4.2.0. Specifically, a bug in mri_em_register was fixed, which was not creating an optimal registration in the instance of your data, and manifesting itself during the mri_watershed step (which uses the talairach_with_skull.lta file created by mri_em_register), where the ouput message was: "GLOBAL region of the brain empty !".

So your options are:

1) update to v4.2.0

or

2) just get the newest mri_em_register, from here: http://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/fixes/

and then the recommended action is to rerun all of your subjects using the new mri_em_register, since it is possible that your other subjects could differ slightly in their results (so to remove that confound, you would need to rerun).

Or you could include the -no-wsgcaatlas option with -skullstrip with your existing v4.0.5, although i can't say whether the downstream mri_em_register will fail for your 19 subjects (the examples you sent seem to be of somewhat low quality in terms of what freesurfer expects. i've attached a tiff of a T1.mgz coronal slice of what we usually see in terms of contrast).

Nick

On Thu, 2009-02-05 at 14:59 -0500, Kathy Zhang wrote:

Hello all,

I've been working with a set of about 200 brains, and in 19 of these I'm getting the error "mri_watershed Error: GLOBAL region of the brain empty !". I've checked the talairach transformation with tkregister2, and in 17 of those cases, the transformations look fine (the two other cases were just bad scans). I'm using freesurfer version 4.0.5 for linux, and have already tried previous suggestions to run a skull strip off orig.mgz and checking the FOV values on T1.mgz and orig.mgz.

Does anyone know what I could do to fix this error or somehow work around it to continue the skull strip?

Thanks, Kathy Zhang _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer