Hi,

I wanted to ask for suggestions on how segmentations and surface reconstruction for CT could be improved in individuals with extensive atrophy.

I noticed that with my data, FreeSurfer outputs (version 7.4.1) have significant errors especially in areas affected by atrophy in that GM is undersegmented. On the other hand, in controls I sometimes see the opposite issue, where parts of the skull are included as GM. 

I have been exploring options like -seg-wlo and -bigventricles to improve the output but wanted to see if anyone has encountered similar issues and had other ideas.

Thanks,

Amelie