Thank you very much for this. It led me exactly to where I needed to look. In regards to this specific subject, it ran through reconstruction with a bad skullstrip but accurate surfaces. We then fixed the skullstrip and reran autorecon 2 & 3, but then the bad surfaces appeared. Today, when I reran mri_em_register (from fs5.0.0) on the fixed skullstrip, I got the talairach.lta file that was way, way off. However, when I use the fs 5.1.0 mri_em_register, I get a talairach.lta that is almost identical to yours. The talairach.lta file seemed to be way off for the rest of these subjects as well. Not sure if anyone else is having problems with the mri_em_register between 5.0.0 and 5.1.0. We will reprocess these in 5.1.0 and hopefully that will fix it.

Jeff

On Thu, Oct 6, 2011 at 4:13 AM, Bruce Fischl <fischl@nmr.mgh.harvard.edu> wrote:

Hi Jeff,

it's hard to sort out what happened as it looks like you invoked recon-all over 50 times for this subject! In any case, the talairach.lta was way, way off, but when I just reran mri_em_register it worked perfectly, so I can't reconstruct what you did. Can you try running recon-all -autorecon2 -autorecon3 and see if it fixes things? Did the skull strip fail for you initially? Maybe em_register and the aseg stuff were run on a bad skull strip?

I'll attach the output of my mri_em_register run (called test.lta). You can compare it to the talairach.lta in that subject's transforms dir using tkmedit or tkregister2.

cheers
Bruce

On Wed, 5 Oct 2011, Jeff Sadino wrote:

Hi Bruce,
Thank you very much for taking a look at this for us. I uploaded it to your
ftp server under the name JeffSadino_050089_S05.tar. About 70 of our
subjects are showing similar problems. It seems to be happening more
frequently in fs5 as compared to fs4. If you can find anything that will
help, that would be great. Thank you again!

Jeff

On Wed, Oct 5, 2011 at 2:41 PM, Bruce Fischl <fischl@nmr.mgh.harvard.edu>
wrote:
Hi jeff
Sorry, this slipped through the cracks. If you upload the dataset we
will take a look
Bruce

On Oct 5, 2011, at 8:32 PM, Jeff Sadino <jsadino.queens@gmail.com>
wrote:

Hello FreeSurfer Experts,
Does anyone have any ideas about how to salvage this data?

Thank you very much!
Jeff

On Sun, Oct 2, 2011 at 5:10 PM, Jeff Sadino
<jsadino.queens@gmail.com> wrote:
Hi Freesurfer Team,
When we tried to process some subjects in FS 5.0.0 through
the cross-sectional step in the longitudinal processing
stream, we got inaccurate surfaces generated. Also, after
doing skullstripping or adding control points and
re-running the cross-sectional reconstruction, inaccurate
pial and white matter surfaces were generated (see
attached pictures). We are not sure how to fix this
problem and this has affected about 70 of our scans.
Several of these subjects we also processed in FS 3 and
FS 4.3.1, and the surfaces were generated fine. Are there
any known solutions to fixing the surfaces in the newer
FreeSurfer? If not, then could we feed the 4.3.1
cross-sectionals into FS 5 to create the base template,
and then use FS 5 to create the longitudinal
reconstructions?

Thank you,
Jeff Sadino

<050089_S05rhlh.png>

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

The information in this e-mail is intended only for the person to whom
it is
addressed. If you believe this e-mail was sent to you in error and the
e-mail
contains patient information, please contact the Partners Compliance
HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you
in error
but does not contain patient information, please contact the sender
and properly
dispose of the e-mail.