Hello,
I’m attempting to generate inner/outer skull and skin surfaces using mne watershed_bem (I don’t have FLASH images). The issue I have is that the outer skull surface is incorrect and looks more like the skin surface, including a large amount of neck. Upon closer inspection, this seems to be due to large chunks of dura included in the brainmask (mostly in the area superior to the interhemispheric fissure), pushing the surface out too far.
I tried the suggestions listed here: Frequently Asked Questions (FAQ) — MNE 1.1.dev0 documentation. Going further back in the FreeSurfer pipeline seemed to work well - I was able to generate accurate brainmask and brain.mgz files by running recon-all -skullstrip with a lower watershed threshold and the -gcut flag. I’ve observed that the -gcut flag is necessary for generating accurate brainmasks with my dataset. My understanding was that this change to brain.mgz should have then improved the surfaces generated when subsequently running mne watershed_bem (also tried mri_watershed in FS), however, when I run it a new brain.mgz file that looked like the original incorrect volume (before using gcut) is instead generated. Is there any way to pass the edited brainmask to mne watershed_bem?
My issue is similar to that posted by another user in the past, but was not resolved (BEM unaffected by edits to brainmask). I also know that mne watershed_bem is a wrapper for mri_watershed in FreeSurfer, and have also posted to the FreeSurfer forum ([Freesurfer] Running mri_watershed with previously generated brainmask) with no response as of yet.
Any advice would be much appreciated.
Thanks in advance,
Danielle