Dear Freesurfer experts,

I am trying to segment the white and pial matter from my MR images but for some reason, the final segmentation always misses some brain areas.  I attached here a nu and brainmask image with the segmentation as an example. The MR image I am using is a MEMPRAGE 0.5mm resolution acquisition from a 3T Siemens TrioTim scanner, possibly with a 64 coil channel. Can you please help me figure out how I could improve the segmentation?

I am using the following command:

recon-all -autorecon1 -autorecon2 -autorecon3 -s hires_test -hires -i high_res.nii

On another note, the full recon-all command takes around four days to finish in an 8GB ram mem computer. I would appreciate if you happen to have any suggestions on how I could run it faster, if this is even possible.

Thanks

Gio.