Dear FreeSurfers,
I am analyzing data from a few patients with quite severe unilateral temporal lobe damage (no or little brain tissue left both medially and laterally). The cortical parcellation in the damaged hemispheres seem - compared to the presumably unaffected contralateral hemisphere and other "normal" brains - all right in some parts, while other labels are misplaced due to the placing of temporal lobe labels which no longer exist. I was looking for some advice as to how to proceed:
Is it advicable to use the labels from the original (recon-all) cortical parcellation that seem well parcellated, and create ROIs manually by drawing in tksurfer/tkmedit the parts of the brain where the labels have been misplaced? Or should I create ROIs manually and don't use the automated cortical parcellation? Or is yet another approach more appropriate, for example changing the input used in the parcellation so that Freesurfer don't expect to find and label temporal lobe structures in the damaged hemisphere?
Thanks!
Håkon Grydeland
Hi Håkon,
could you give us more detail about what you are trying to achieve?
cheers, Bruce On Fri, 4 Apr 2008 hakon.grydeland@psykologi.uio.no wrote:
Dear FreeSurfers,
I am analyzing data from a few patients with quite severe unilateral temporal lobe damage (no or little brain tissue left both medially and laterally). The cortical parcellation in the damaged hemispheres seem - compared to the presumably unaffected contralateral hemisphere and other "normal" brains - all right in some parts, while other labels are misplaced due to the placing of temporal lobe labels which no longer exist. I was looking for some advice as to how to proceed:
Is it advicable to use the labels from the original (recon-all) cortical parcellation that seem well parcellated, and create ROIs manually by drawing in tksurfer/tkmedit the parts of the brain where the labels have been misplaced? Or should I create ROIs manually and don't use the automated cortical parcellation? Or is yet another approach more appropriate, for example changing the input used in the parcellation so that Freesurfer don't expect to find and label temporal lobe structures in the damaged hemisphere?
Thanks!
Håkon Grydeland
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi Bruce,
I am trying to parcellate the cortex following the normal processing stream (recon-all) in order to get estimations of cortical thickness and volume in the cortical labels/annotations (?h.aparc.annot).
As you can see from the attached files the automatic cortical parcellation works fairly well, but there seem to be some errors in the parcellation due to the discrepancy between the labels used and the "corresponding" brain tissue (i.e. the labels for the anterior temporal lobe are superfluous because of the loss of brain tissue in this area, but parts of the cortex is still labeled with these labels because of the probabilistic atlas). In addition there also seem to be displacements of some labels for which the brain tissue is intact.
I am therefore wondering what would be the best way of doing the cortical parcellation in order to label the cortex outside of the temporal lobe in a precise manner; modifying the automatic cortical parcellation, by doing some of it manually, etc.? I hope this made things clearer!??
Thanks again!
Håkon
Hi Håkon,
could you give us more detail about what you are trying to achieve?
cheers, Bruce On Fri, 4 Apr 2008 hakon.grydeland@psykologi.uio.no wrote:
Dear FreeSurfers, I am analyzing data from a few patients with quite severe unilateral
temporal lobe damage (no or little brain tissue left both medially and laterally). The cortical parcellation in the damaged hemispheres seem - compared to the presumably unaffected contralateral hemisphere and other
"normal" brains - all right in some parts, while other labels are
misplaced due to the placing of temporal lobe labels which no longer exist. I was looking for some advice as to how to proceed:
Is it advicable to use the labels from the original (recon-all)
cortical
parcellation that seem well parcellated, and create ROIs manually by
drawing in tksurfer/tkmedit the parts of the brain where the labels have
been misplaced? Or should I create ROIs manually and don't use the
automated cortical parcellation? Or is yet another approach more appropriate, for example changing the input used in the parcellation so that Freesurfer don't expect to find and label temporal lobe structures in
the damaged hemisphere? Thanks! Håkon Grydeland _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi Håkon,
editing will probably be a *lot* easier than generating a parcellation from scratch, and probably more reproducible as well, but of course it depends on your background and experience. If you've done a lot of manual parcellating maybe you can do better.
cheers, Bruce
On Fri, 4 Apr 2008 hakon.grydeland@psykologi.uio.no wrote:
Hi Bruce,
I am trying to parcellate the cortex following the normal processing stream (recon-all) in order to get estimations of cortical thickness and volume in the cortical labels/annotations (?h.aparc.annot).
As you can see from the attached files the automatic cortical parcellation works fairly well, but there seem to be some errors in the parcellation due to the discrepancy between the labels used and the "corresponding" brain tissue (i.e. the labels for the anterior temporal lobe are superfluous because of the loss of brain tissue in this area, but parts of the cortex is still labeled with these labels because of the probabilistic atlas). In addition there also seem to be displacements of some labels for which the brain tissue is intact.
I am therefore wondering what would be the best way of doing the cortical parcellation in order to label the cortex outside of the temporal lobe in a precise manner; modifying the automatic cortical parcellation, by doing some of it manually, etc.? I hope this made things clearer!??
Thanks again!
Håkon
Hi Håkon,
could you give us more detail about what you are trying to achieve?
cheers, Bruce On Fri, 4 Apr 2008 hakon.grydeland@psykologi.uio.no wrote:
Dear FreeSurfers, I am analyzing data from a few patients with quite severe unilateral
temporal lobe damage (no or little brain tissue left both medially and laterally). The cortical parcellation in the damaged hemispheres seem - compared to the presumably unaffected contralateral hemisphere and other
"normal" brains - all right in some parts, while other labels are
misplaced due to the placing of temporal lobe labels which no longer exist. I was looking for some advice as to how to proceed:
Is it advicable to use the labels from the original (recon-all)
cortical
parcellation that seem well parcellated, and create ROIs manually by
drawing in tksurfer/tkmedit the parts of the brain where the labels have
been misplaced? Or should I create ROIs manually and don't use the
automated cortical parcellation? Or is yet another approach more appropriate, for example changing the input used in the parcellation so that Freesurfer don't expect to find and label temporal lobe structures in
the damaged hemisphere? Thanks! Håkon Grydeland _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Okey, thanks a lot!
I have another question though: Will Freesurfer (in the automatic cortical parcellation) always try to parcellate/label all of the cortex? I was wondering if a solution to the problem could be to either change and rename the ?h.curvature.buckner40.filled.deskian_killiany.2007-06-20.gsc-file (by deleting the labels in the temporal lobe, if this is possible?) or to make a new atlas based on other scans we have using the parcellations Freesurfer generate in the automatic parcellation, but omit the temporal lobe parcellations in the new annotation files (of course the first alternative would be the easiest), and then re-run -autorecon3 from mris_ca_label. This would just turn the problem around though if the automatic parcellation labels posterior parts of the temporal lobe which are still present in the patients by extending the parietal and/or occipital labels, hence my initial question.
Thanks once again!
Håkon
Hi Håkon,
editing will probably be a *lot* easier than generating a parcellation
from
scratch, and probably more reproducible as well, but of course it
depends
on your background and experience. If you've done a lot of manual
parcellating maybe you can do better.
cheers, Bruce
On Fri, 4 Apr 2008 hakon.grydeland@psykologi.uio.no wrote:
Hi Bruce, I am trying to parcellate the cortex following the normal processing
stream (recon-all) in order to get estimations of cortical thickness and
volume in the cortical labels/annotations (?h.aparc.annot). As you can see from the attached files the automatic cortical parcellation works fairly well, but there seem to be some errors in the parcellation
due to the discrepancy between the labels used and the "corresponding" brain tissue (i.e. the labels for the anterior temporal lobe are superfluous because of the loss of brain tissue in this area, but parts of
the cortex is still labeled with these labels because of the probabilistic atlas). In addition there also seem to be displacements of some labels
for
which the brain tissue is intact. I am therefore wondering what would be the best way of doing the
cortical
parcellation in order to label the cortex outside of the temporal lobe
in
a precise manner; modifying the automatic cortical parcellation, by
doing
some of it manually, etc.? I hope this made things clearer!?? Thanks again! Håkon
Hi Håkon, could you give us more detail about what you are trying to achieve?
cheers,
Bruce On Fri, 4 Apr 2008 hakon.grydeland@psykologi.uio.no wrote:
Dear FreeSurfers, I am analyzing data from a few patients with quite severe unilateral
temporal lobe damage (no or little brain tissue left both medially and
laterally). The cortical parcellation in the damaged hemispheres seem - compared to the presumably unaffected contralateral hemisphere and other
"normal" brains - all right in some parts, while other labels are
misplaced due to the placing of temporal lobe labels which no longer
exist. I was looking for some advice as to how to proceed:
Is it advicable to use the labels from the original (recon-all)
cortical
parcellation that seem well parcellated, and create ROIs manually by
drawing in tksurfer/tkmedit the parts of the brain where the labels
have
been misplaced? Or should I create ROIs manually and don't use the
automated cortical parcellation? Or is yet another approach more
appropriate, for example changing the input used in the parcellation so that Freesurfer don't expect to find and label temporal lobe structures in
the damaged hemisphere? Thanks! Håkon Grydeland _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
no, sorry, it's not that easy. We really expect the whole hemi to be there and there's no easy way to remove part of it
Bruce On Mon, 7 Apr 2008 hakon.grydeland@psykologi.uio.no wrote:
Okey, thanks a lot!
I have another question though: Will Freesurfer (in the automatic cortical parcellation) always try to parcellate/label all of the cortex? I was wondering if a solution to the problem could be to either change and rename the ?h.curvature.buckner40.filled.deskian_killiany.2007-06-20.gsc-file (by deleting the labels in the temporal lobe, if this is possible?) or to make a new atlas based on other scans we have using the parcellations Freesurfer generate in the automatic parcellation, but omit the temporal lobe parcellations in the new annotation files (of course the first alternative would be the easiest), and then re-run -autorecon3 from mris_ca_label. This would just turn the problem around though if the automatic parcellation labels posterior parts of the temporal lobe which are still present in the patients by extending the parietal and/or occipital labels, hence my initial question.
Thanks once again!
Håkon
Hi Håkon,
editing will probably be a *lot* easier than generating a parcellation
from
scratch, and probably more reproducible as well, but of course it
depends
on your background and experience. If you've done a lot of manual
parcellating maybe you can do better.
cheers, Bruce
On Fri, 4 Apr 2008 hakon.grydeland@psykologi.uio.no wrote:
Hi Bruce, I am trying to parcellate the cortex following the normal processing
stream (recon-all) in order to get estimations of cortical thickness and
volume in the cortical labels/annotations (?h.aparc.annot). As you can see from the attached files the automatic cortical parcellation works fairly well, but there seem to be some errors in the parcellation
due to the discrepancy between the labels used and the "corresponding" brain tissue (i.e. the labels for the anterior temporal lobe are superfluous because of the loss of brain tissue in this area, but parts of
the cortex is still labeled with these labels because of the probabilistic atlas). In addition there also seem to be displacements of some labels
for
which the brain tissue is intact. I am therefore wondering what would be the best way of doing the
cortical
parcellation in order to label the cortex outside of the temporal lobe
in
a precise manner; modifying the automatic cortical parcellation, by
doing
some of it manually, etc.? I hope this made things clearer!?? Thanks again! Håkon
Hi Håkon, could you give us more detail about what you are trying to achieve?
cheers,
Bruce On Fri, 4 Apr 2008 hakon.grydeland@psykologi.uio.no wrote:
Dear FreeSurfers, I am analyzing data from a few patients with quite severe unilateral
temporal lobe damage (no or little brain tissue left both medially and
laterally). The cortical parcellation in the damaged hemispheres seem - compared to the presumably unaffected contralateral hemisphere and other
"normal" brains - all right in some parts, while other labels are
misplaced due to the placing of temporal lobe labels which no longer
exist. I was looking for some advice as to how to proceed:
Is it advicable to use the labels from the original (recon-all)
cortical
parcellation that seem well parcellated, and create ROIs manually by
drawing in tksurfer/tkmedit the parts of the brain where the labels
have
been misplaced? Or should I create ROIs manually and don't use the
automated cortical parcellation? Or is yet another approach more
appropriate, for example changing the input used in the parcellation so that Freesurfer don't expect to find and label temporal lobe structures in
the damaged hemisphere? Thanks! Håkon Grydeland _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
hakon.grydeland＠psykologi.uio.no