We have a large number of scans with excellent, visually inspected and manually trimmed brain masks, so we would like to use these masks to assist the FreeSurfer processing. One way we have experimented with to achieve this is to clip the T1 to remove anything non-brain before importing it into the subject directory. In using brain-clipped T1 images as a starting point for FreeSurfer we notice that the Talairach transform checking is quite far off, scaled badly and rotated downward 10+ degrees in pitch. I assume that this is due to the lack of similarity to the moving image, which includes skull. However, even if we don't modify this transform, the final parcellation looks quite good, and very, very similar to that from using a standard, non-clipped T1 as input. Is there any reason not to use a clipped T1 as input into FreeSurfer? Is the Talairach-based information used for segmentation or parcellation, and are we potentially creating problems? Or would it be better to simply use our mask to trim up the images after autorecon1 (coregistering them into the same image space and clipping them with a custom routine)?
Ron Pierson
Department of Psychiatry
Carver College of Medicine
W-278 GH, 200 Hawkins Dr
Iowa City, IA 52242
Hi Ron,
we do use the talairach for things like estimating icv, so that would concern me a bit. I don't think there is a straightforward way to insert the skull stripped image into our stream though. Maybe Nick has a better idea?
Bruce
On Thu, 28 Jan 2010, Pierson, Ronald K wrote:
We have a large number of scans with excellent, visually inspected and manually trimmed brain masks, so we would like to use these masks to assist the FreeSurfer processing. One way we have experimented with to achieve this is to clip the T1 to remove anything non-brain before importing it into the subject directory. In using brain-clipped T1 images as a starting point for FreeSurfer we notice that the Talairach transform checking is quite far off, scaled badly and rotated downward 10+ degrees in pitch. I assume that this is due to the lack of similarity to the moving image, which includes skull. However, even if we don't modify this transform, the final parcellation looks quite good, and very, very similar to that from using a standard, non-clipped T1 as input. Is there any reason not to use a clipped T1 as input into FreeSurfer? Is the Talairach-based information used for segmentation or parcellation, and are we potentially creating problems? Or would it be better to simply use our mask to trim up the images after autorecon1 (coregistering them into the same image space and clipping them with a custom routine)?
Ron Pierson
Department of Psychiatry
Carver College of Medicine
W-278 GH, 200 Hawkins Dr
Iowa City, IA 52242
you could break it apart:
recon-all -s subjid -autorecon1 -noskullstrip
cp <your skull stripped image> brainmask.auto.mgz cp brainmask.auto.mgz brainmask.mgz
recon-all -s subjid -autorecon2 -autorecon3
n.
On Thu, 2010-01-28 at 15:31 -0500, Bruce Fischl wrote:
Hi Ron,
we do use the talairach for things like estimating icv, so that would concern me a bit. I don't think there is a straightforward way to insert the skull stripped image into our stream though. Maybe Nick has a better idea?
Bruce
On Thu, 28 Jan 2010, Pierson, Ronald K wrote:
We have a large number of scans with excellent, visually inspected and manually trimmed brain masks, so we would like to use these masks to assist the FreeSurfer processing. One way we have experimented with to achieve this is to clip the T1 to remove anything non-brain before importing it into the subject directory. In using brain-clipped T1 images as a starting point for FreeSurfer we notice that the Talairach transform checking is quite far off, scaled badly and rotated downward 10+ degrees in pitch. I assume that this is due to the lack of similarity to the moving image, which includes skull. However, even if we don't modify this transform, the final parcellation looks quite good, and very, very similar to that from using a standard, non-clipped T1 as input. Is there any reason not to use a clipped T1 as input into FreeSurfer? Is the Talairach-based information used for segmentation or parcellation, and are we potentially creating problems? Or would it be better to simply use our mask to trim up the images after autorecon1 (coregistering them into the same image space and clipping them with a custom routine)?
Ron Pierson
Department of Psychiatry
Carver College of Medicine
W-278 GH, 200 Hawkins Dr
Iowa City, IA 52242
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
i forgot to add: dont use the skull-stripped image as input. your's just gets inserted after -autorecon1 to replace brainmask.<auto>.mgz
n. On Thu, 2010-01-28 at 15:59 -0500, Nick Schmansky wrote:
you could break it apart:
recon-all -s subjid -autorecon1 -noskullstrip
cp <your skull stripped image> brainmask.auto.mgz cp brainmask.auto.mgz brainmask.mgz
recon-all -s subjid -autorecon2 -autorecon3
n.
On Thu, 2010-01-28 at 15:31 -0500, Bruce Fischl wrote:
Hi Ron,
we do use the talairach for things like estimating icv, so that would concern me a bit. I don't think there is a straightforward way to insert the skull stripped image into our stream though. Maybe Nick has a better idea?
Bruce
On Thu, 28 Jan 2010, Pierson, Ronald K wrote:
We have a large number of scans with excellent, visually inspected and manually trimmed brain masks, so we would like to use these masks to assist the FreeSurfer processing. One way we have experimented with to achieve this is to clip the T1 to remove anything non-brain before importing it into the subject directory. In using brain-clipped T1 images as a starting point for FreeSurfer we notice that the Talairach transform checking is quite far off, scaled badly and rotated downward 10+ degrees in pitch. I assume that this is due to the lack of similarity to the moving image, which includes skull. However, even if we don't modify this transform, the final parcellation looks quite good, and very, very similar to that from using a standard, non-clipped T1 as input. Is there any reason not to use a clipped T1 as input into FreeSurfer? Is the Talairach-based information used for segmentation or parcellation, and are we potentially creating problems? Or would it be better to simply use our mask to trim up the images after autorecon1 (coregistering them into the same image space and clipping them with a custom routine)?
Ron Pierson
Department of Psychiatry
Carver College of Medicine
W-278 GH, 200 Hawkins Dr
Iowa City, IA 52242
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Thanks! Yes, that makes perfect sense - looking at:
http://surfer.nmr.mgh.harvard.edu/fswiki/ReconAllDevTable
Ron
-----Original Message----- From: Nick Schmansky [mailto:nicks@nmr.mgh.harvard.edu] Sent: Thursday, January 28, 2010 3:04 PM To: Bruce Fischl Cc: Pierson, Ronald K; freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Using a clipped image
i forgot to add: dont use the skull-stripped image as input. your's just gets inserted after -autorecon1 to replace brainmask.<auto>.mgz
n. On Thu, 2010-01-28 at 15:59 -0500, Nick Schmansky wrote:
you could break it apart:
recon-all -s subjid -autorecon1 -noskullstrip
cp <your skull stripped image> brainmask.auto.mgz cp brainmask.auto.mgz brainmask.mgz
recon-all -s subjid -autorecon2 -autorecon3
n.
On Thu, 2010-01-28 at 15:31 -0500, Bruce Fischl wrote:
Hi Ron,
we do use the talairach for things like estimating icv, so that
would
concern me a bit. I don't think there is a straightforward way to
insert
the skull stripped image into our stream though. Maybe Nick has a
better
idea?
Bruce
On Thu, 28 Jan 2010, Pierson, Ronald K wrote:
We have a large number of scans with excellent, visually inspected
and
manually trimmed brain masks, so we would like to use these masks
to
assist the FreeSurfer processing. One way we have experimented
with to
achieve this is to clip the T1 to remove anything non-brain before importing it into the subject directory. In using brain-clipped
T1
images as a starting point for FreeSurfer we notice that the
Talairach
transform checking is quite far off, scaled badly and rotated
downward
10+ degrees in pitch. I assume that this is due to the lack of similarity to the moving image, which includes skull. However,
even if
we don't modify this transform, the final parcellation looks quite
good,
and very, very similar to that from using a standard, non-clipped
T1 as
input. Is there any reason not to use a clipped T1 as input into FreeSurfer? Is the Talairach-based information used for
segmentation or
parcellation, and are we potentially creating problems? Or would
it be
better to simply use our mask to trim up the images after
autorecon1
(coregistering them into the same image space and clipping them
with a
custom routine)?
Ron Pierson
Department of Psychiatry
Carver College of Medicine
W-278 GH, 200 Hawkins Dr
Iowa City, IA 52242
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
Nick Schmansky -
Pierson, Ronald K