Problem with cat12 - creates cube shaped head mask

I'm having a problem with CAT12 segmentation, where the head mask comes out as a cube. The brain inside is fine but I can't use the head for source reconstruction.

Interestingly, this has only happened for 2 out of 10+ MRI scans with the same acquisition settings. I have followed this up with the MRI team and nothing seems to be wrong with the acquisition - the tech was able to use CAT12 without an issue (outside of brainstorm).

I haven't been able to understand why this is happening. The only thing I note is that the scan is a bit dark, but the contrast seems to be fine.

Do you have any tips?
Thank you!

The head surface is generated by Brainstorm, not by CAT12.

First, check that the outputs of CAT12 are correct: Open the cortex surface and the volume atlases to make sure the segmentation worked correctly.

Then try to recompute a different head surface from the MRI, following the instructions on that page: