External Email - Use Caution
Thanks, Niels.
It is definitely a matter of hyperintensity of voxels close to the lateral
boundary, particularly in the anterior region (which has a fuzzy boundary with
the white matter in first place). Whether these hyperintensities are due to MS
pathology, that I do not know (I’m really not the person to tell…).
These could be cleaned up with postprocessing (e.g., filling holes, or using a
Markov Random Field), but the impact on the volumes would be negligible.
Cheers,
/Eugenio
--
Juan Eugenio Iglesias
ERC Senior Research Fellow
Centre for Medical Image Computing (CMIC)
University College London
http://www.jeiglesias.com
http://cmictig.cs.ucl.ac.uk/
From: Niels Bergsland
Date: Wednesday, 10 October 2018 at 15:42
To: "Iglesias Gonzalez, Eugenio"
Subject: Re: [Freesurfer] Disjointed thalamic nuclei segmentation
Hi Eugenio,
Thanks for the lightning fast response! :)
I've attached them here. It looks like it might actually be the opposite in
this case in that the voxels here are darker. Some of the cases are MS patients
and so it's conceivable that they have thalamic lesions. Admittedly though, I
have not really spent time in the past looking for thalamic lesions on
T1-weighted images and this hasn't been an issue in the past since the thalamic
segmentation in the aseg doesn't end up being affected by these voxels with
lower intensities. If you would like to look at some other cases or need other
outputs, just let me know and I'm happy to send them along!
Thanks again,
Niels
On Wed, Oct 10, 2018 at 4:30 PM Iglesias Gonzalez, Eugenio
mailto:e.igles...@ucl.ac.uk>> wrote:
External Email - Use Caution
Thanks, Niels.
I suspect the problem is that, for some reason, such voxels have much brighter
intensity than their surroundings, and are classified as white matter /
reticular nucleus. But that shouldn’t be happening that far from the lateral
boundary of the thalamus…
Can you please send us norm.mgz and the thalamic segmentation file, so I can
take a look?
Cheers,
/Eugenio
--
Juan Eugenio Iglesias
ERC Senior Research Fellow
Centre for Medical Image Computing (CMIC)
University College London
http://www.jeiglesias.com
http://cmictig.cs.ucl.ac.uk/
From:
mailto:freesurfer-boun...@nmr.mgh.harvard.edu>>
on behalf of Niels Bergsland mailto:theni...@gmail.com>>
Reply-To: Freesurfer support list
mailto:freesurfer@nmr.mgh.harvard.edu>>
Date: Wednesday, 10 October 2018 at 15:22
To: Freesurfer Mailing List
mailto:freesurfer@nmr.mgh.harvard.edu>>
Subject: [Freesurfer] Disjointed thalamic nuclei segmentation
External Email - Use Caution
Hi Eugenio,
Thank you again for the fantastic thalamic nuclei segmentation tool!
I'm going through and QC'ing a batch of subjects that were processed through
the pipeline. Data was originally processed with freesurfer-6.0.0 and then
processed through the thalamic nuclei stream (dev-20180818-e30e6f9).
I have found that for some cases, there are some isolated patches of voxels
that are disconnected and seemingly in the wrong place. I've noticed it
primarily for the MDm and my impression is that it tends to happen more in the
right hemisphere. I've attached a sample image to show you what I mean. The red
cross has been placed on a set of these voxels. It is not the case that they
are connected in 3D.
Any input is appreciated and thanks again! If it can be useful, I'm happy to
upload the data
-Niels
___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu<mailto: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
