After running &quot;recon-all -autorecon1&quot; I saw that brain.mgz was missing some large areas in the cerebellum.&nbsp; <br>
I ran &quot;recon-all -skullstrip -wsthreshold 35&quot; and then checked my work with &quot;tkmedit subjid brainmask.auto.mgz -aux T1.mgz&quot;<br>
The skullstrip looked very good, so I went on to run autorecon2.&nbsp;
brain.mgz was never updated and the output of autorecon2 was
horrendous.&nbsp; <br>
I am going to use mri_mask to generate a new brain.mgz using T1.mgz and
brainmask.auto.mgz.&nbsp; Is this an appropriate solution to my
problem?&nbsp; The document fstutorial/skullstriptutorial, which I used
as my guide, indicates tht running recon-all with the -skullstrip flag
will generate a new brain.mgz but this did not occur when I used the
command.&nbsp; The autorecon help file correctly indicates that
recon-all -skullstrip generates brainmask.auto.mgz and brainmask.mgz
but lacks explanation of the command's various modes.&nbsp; It appears
that the generation of brain.mgz is somehow lost between autorecon1 and
autorecon2, at least in my experience. &nbsp; <br>
<br>
FYI I am using freesurfer stable release 3.0.1 on a pentium 4 running Fedora Core 4<br>
<br>
-Tom Bayer<br>