Dear Daniel,

For the "outer" boundary of your ROI, I would try the betsurf from FSL for the inner skull surface estimation. This should be more precise than using brainmask.mgz.

https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/BET/UserGuide

Antonin

You could remove the brain (as found in aseg.mgz) from the 
brainmask.mgz. This will leave some dura. It is not an elegant or 
accurate solution

On 03/06/2017 02:24 PM, Albrecht, Daniel S. wrote:
> Hello,
>
> We have a PET dataset where we expect to see pathogenic changes in the 
> meninges. We'd like to extract the meningeal signal in subject space, 
> and were hoping to come up with an automated strategy to create 
> meningeal ROIs, containing no brain or scalp.
> So my question is: can you suggest any way within the FreeSurfer 
> pipeline that we can create a meningeal-only ROI, or alternatively an 
> ROI containing both skull and meninges, but still no scalp or brain?
>
> Your help is greatly appreciated!
>
> Daniel S. Albrecht, PhD
> Research Fellow in Radiology
> Martinos Center for Biomedical Imaging