Is it a flag after recon-all? Could you show me an example of how to set gm T1.mgz values?
Thank you very much!
Victor
On Tue, Aug 13, 2013 at 3:55 PM, Bruce Fischl
<fischl@nmr.mgh.harvard.edu> wrote:
Hi Victor
yes, look at the expert options. You can set the range of allowable wm and gm values (and csf for that matter).
cheers
Bruce
On Tue, 13 Aug 2013, Victor Kovac wrote:
Hi Bruce,
I was using v5.1 but switched to v5.3 for these troublesome cases. The
biggest issue is that this is a longitudinal study and these scans are from
patients' 4th visits. Looking in the asegstats tables from all 4 visits, it
would appear that the patients having this issue have had a ~20% increase in
cortical WM since their 3rd visit last year, after showing a steady decline
previously.
There isn't great GM/WM contrast, so I added control points to this subject.
That reset the T1.mgz value of voxels that I would consider to be WM to 109
or higher, yet I still see nearly all GM with T1.mgz values in the 80's and
90's segmented as WM, which is why I am confused.
From what I can see, for this brain (MPS3008-4) only voxels with a T1.mgz of
less than 80 are being segmented as GM. Is there a way for me to manually
plug in the voxel intensity ranges for GM and WM and re-create the surfaces?
Thank you very much for your help with this!
Victor
On Tue, Aug 13, 2013 at 3:29 PM, Bruce Fischl <fischl@nmr.mgh.harvard.edu>
wrote:
never mind, I see the problem. I think this is due to the lack
of gray/white contrast in this subject. I'll try some control
points, but it may be that it will need some expert opts to
specify the white and pial intensity ranges
On Mon, 12 Aug 2013, Victor Kovac wrote:
Dear Bruce,
I have uploaded the subject with unsegmented
ventricles (MPS3025-4) via File
Drop. This subject has an artifact from a
ventricular shunt in his left
hemisphere, and we are only using data from his
right hemisphere.
I have also uploaded a subject (image attached)
whose cortical GM is being
segmented as WM, and whose pial boundary is extended
past the cortex. This
is a recurrent issue with the atrophied brains I am
working with. I added
control points to bright white matter voxels in
order to re-threshold WM
voxel intensity and hoped that re-running
autorecon-cp -autorecon3 would fix
the issue, but it seems to have done nothing.
Could you take a look at this subject and tell me if
you know there is a way
to get FreeSurfer to take my re-thresholding
attempts and recreate the pial
and white matter boundaries?
Please let me know if there was an issue with my
upload, and, again, thank
you for your help!
Victor
On Fri, Aug 9, 2013 at 2:28 PM, Bruce Fischl
<fischl@nmr.mgh.harvard.edu>
wrote:
hmmm, you could manually correct it but if you
upload that
dataset I'll take a look as well
On Fri, 9 Aug 2013, Victor Kovac wrote:
Dear Bruce,
As I have been looking through brains I
have run
recon-all on with v5.3, I'm
noticing that there are many cases of
lateral
ventricles being left
unsegmented in subjects with dilated
ventricles.
An image of one case of unsegmented
ventricles is
attached, where a subject
has an artifact from his shunt which may
be causing
the error. I have
manually adjusted this subject's
talairach to fix
its talairach failure.
Should I be using the attached script
"tkmedit-lv.sh" to fill the
ventricles, or is there a better method?
Thank you very much for your help.
Victor
On Fri, Aug 2, 2013 at 1:13 PM, Bruce
Fischl
<fischl@nmr.mgh.harvard.edu>
wrote:
Hi Victor,
yes, 5.3 is most of what you need.
I think I
added a couple of
control points as well, but they
should be in
the subject dir i
sent you.
cheers
Bruce
On Fri, 2 Aug 2013, Victor Kovac
wrote:
Bruce,
Thank you so much for
analyzing this
subject -- the
segmentation looks much
better! Was it simply
running the
subject with v5.3,
or did you add any
flags to the recon?
As you mentioned, the pial
and white
matter surfaces
extend further than
they should, but I have
found adding
control points
to white matter near the
cortical surface has helped
a lot in
similar cases.
I have two other subjects
whose
ventricles were left
unsegmented, but have
less atrophy than the
subject I sent
you. I tried
filling the ventricles
using the attached script,
then ran
recon with
-autorecon2-wm -autorecon3
flags, but it did not have
any effect.
Would you
recommend just running the
subject with v5.3?
Thank you!
Victor
On Thu, Aug 1, 2013 at 10:32
AM, Bruce
Fischl
<fischl@nmr.mgh.harvard.edu>
wrote:
Hi Victor
the 5.3 recon finished
and you can
grab the
subject from:
http://gate.nmr.mgh.harvard.edu/filedrop2/?p=if8d3r4
I think the ventricles
are *much*
better. The
white surface was
incorrect in anterior
temporal
regions due to
lack of
myelination, but that
is expected.
You might
be able to improve
them, but there is
little enough
contrast that
I'm not sure how
good they will be
cheers
Bruce
On Tue, 30 Jul 2013,
Victor Kovac
wrote:
Dear Freesurfer
experts,
I am running recon-all
on very
atrophied
brains with mixed
results. Attached
is an image of a brain
where
segmentation has
gone awry.
There is no message
indicating talairach
failure, and
the
segmentation appears
to need WM and
pial edits as well as
the
ventricles to be
filled, so I am
not sure how to
proceed.
If you have any
recommendations as
to how to
fix these
errors, I would very
much appreciate it --
these data
are past
their deadline,
and I am at a
loss.
Thank you!
Sincerely,
Victor
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.