Hi everyone,

In some cases I get a really bad subcortical segmentation and I was wondering what the reason might be and how I could fix this. Attached is a screenshot showing the nu_noneck.mgz with the aseg.mgz overlaid.

Viewing this screenshot I would say, that during mri_remove_neck the “possible brain” is not correct, therefore more tissue is erased and the talairach_with_skull.lta is wrong. Though it is not written in the ReconAllDevTable I would guess, that the bad talairach_with_skull.lta is used during the following stages which affects the aseg.mgz.

1. What does the “possible brain” during mri_remove_neck refer to?

2. Could this simply be fixed by using a greater radius for mri_remove_neck or is most probably something else terribly wrong?

Thanks in advance,

Falk