Hi, everone. The skull strip quailtiy from the reconall freesurfer was bad. I know I can manually revise the watershed parameters, but it will take a whil since I had over 30 patients. Therefore, is it possible to use the good quaility of brian mask in the recon-all process so that I can skilp the skull strip process?

Many thanks

At least in recent versions, you can pass -noskullstrip -xmask /path/to/good/mask.nii.gz.

1 Like

Thanks for your reply. Then If I use my own mask.nii.gz with using -noskullstrip -xmask /path/to/good/mask.nii.gz , then will the reconall starts the first step of segmentation (autorecon1) using the brain mask I gave?

You need to pass -noskullstrip when you run -all or -autorecon1. If you usually do:

recon-all -s sub-01 -i sub-01/anat/sub-01_T1w.nii.gz -all

Then you would instead write:

recon-all -s sub-01 -i sub-01/anat/sub-01_T1w.nii.gz \
    -xmask sub-01/anat/sub-01_mask.nii.gz -all -noskullstrip

(-noskullstrip needs to come after -all, but -xmask can come anywhere.)

1 Like

Thanks for your suggestion!

I’m using the freesurfer (7.4). And my main goal is to get a good segmentation quality for brainstem, cerebellum and thalamus.

With the freeusrfer using -xamsk option, I found an issue where recon-all expects my brain mask placed to the result directory I set.

So, if I place my brain mask to the result directory before run the recon -all, the script won’ t run since my customize brain mask already existed in the result directory. It seems like the recon-all pipeline is having trouble with a pre-existing brainmask.mgz which I placed before running the recon -all.

In this case, would you recommend to run only autoren1 stage first, and replace the brain.mgz file with my customize brain mask. And the, run the rest of stage 2 and 3?

Many thanks.

That would make sense, yes.