well, something is really wrong. Did you happen to change the voxel type of brain.finalsurfs.mgz? Can you run mri_info on it and send us the results? If not, then upload the tarred and gzipped subject dir and I'll take a look. I don't think it will be hard to fix
On Thu, 28 Mar 2013, preciado@nmr.mgh.harvard.edu wrote:
A) Log attached. B) That was my next step but I wanted to check first to see if there was something obvious I was missing.
I actually ran "recon-all -autorecon-pial -subjid subject_name" twice. I thought the first time, that it was the result of some process failing/not completing along the way without entirely crashing freesurfer. I ran it again in the hopes that it was some crazy fluke but the result is exactly the same.
-Ronny
eek, that's pretty bad. No idea why that would happen. Can you send us the recon-all.log? If you rerun with the original brain.finalsurfs.mgz does it restore good surfaces? On Thu, 28 Mar 2013, preciado@nmr.mgh.harvard.edu wrote:
Hello Freesurfers,
I ran recon-all on a subject's structural data like so: recon-all -autorecon-all -subjid subect_name
The resulting pial surface looked pretty good except that there were a few voxels of dura and cerebellum that were included in the pial surface. not a big deal. I deleted some voxels of dura from brainmask.mgz and deleted voxels of from the cerebellum by creating and saving a copy of brain.finalsurfs.mgz. /subject_name/mri/ cp brain.finalsurfs.mgz brain.finalsurfs.manedit.mgz
Then I ran recon all to regenerate the pial surface: recon-all -autorecon-pial -subjid subject_name
The results from the pial surface redrawing are now totally wrong and completely underestimated. The weird thing is that it looked perfectly fine before and removing some problematic voxels caused this result.
Any insights as to how/why this occurred and how to fix it? Images of the AFTER picture attached.
Thanks! -Ronny
mri_info for both attached.
Interestingly enough, brain.finalsurfs.manedit.mgz is significantly larger in size. If it was a copy of the other, shouldn't it be the same size? Also, brain.finalsurfs.manedit.mgz shows a time/date stamp from the last time I edited it and brain.finalsurfs.mgz has a time/date stamp from the last time I executed recon-all -autorecon-pial...
-rw-rw-r-- 1 preciado birn 1416877 Mar 26 22:23 brain.finalsurfs.manedit.mgz -rw-rw-r-- 1 preciado birn 16605 Mar 27 09:39 brain.finalsurfs.mgz
A more basic question, I am assuming that -autorecon-pial defaults to using brain.finalsurfs.manedit.mgz and then proceeds to use brain.finalsurfs.mgz if there is no manual edit volume, is that correct?
-Ronny
well, something is really wrong. Did you happen to change the voxel type of brain.finalsurfs.mgz? Can you run mri_info on it and send us the results? If not, then upload the tarred and gzipped subject dir and I'll take a look. I don't think it will be hard to fix
On Thu, 28 Mar 2013, preciado@nmr.mgh.harvard.edu wrote:
A) Log attached. B) That was my next step but I wanted to check first to see if there was something obvious I was missing.
I actually ran "recon-all -autorecon-pial -subjid subject_name" twice. I thought the first time, that it was the result of some process failing/not completing along the way without entirely crashing freesurfer. I ran it again in the hopes that it was some crazy fluke but the result is exactly the same.
-Ronny
eek, that's pretty bad. No idea why that would happen. Can you send us the recon-all.log? If you rerun with the original brain.finalsurfs.mgz does it restore good surfaces? On Thu, 28 Mar 2013, preciado@nmr.mgh.harvard.edu wrote:
Hello Freesurfers,
I ran recon-all on a subject's structural data like so: recon-all -autorecon-all -subjid subect_name
The resulting pial surface looked pretty good except that there were a few voxels of dura and cerebellum that were included in the pial surface. not a big deal. I deleted some voxels of dura from brainmask.mgz and deleted voxels of from the cerebellum by creating and saving a copy of brain.finalsurfs.mgz. /subject_name/mri/ cp brain.finalsurfs.mgz brain.finalsurfs.manedit.mgz
Then I ran recon all to regenerate the pial surface: recon-all -autorecon-pial -subjid subject_name
The results from the pial surface redrawing are now totally wrong and completely underestimated. The weird thing is that it looked perfectly fine before and removing some problematic voxels caused this result.
Any insights as to how/why this occurred and how to fix it? Images of the AFTER picture attached.
Thanks! -Ronny
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
preciado＠nmr.mgh.harvard.edu