Dear Octavian,
the current philosophy with wm.mgz edits is not to directly affect voxel values in original T1 image.
The wm.mgz serves as starting point for the white surface estimation. The ?h.white surface is estimated according gradient in voxel values in brain.finalsurfs.mgz (which is, basically, preprocessed T1 image) using basically preprocessed surfaces obtained from wm.mgz (?h.orig) as a starting position where to search for intensity gradient. The editing of wm.mgz helps to find out the GM/WM itensity gradient and properly place ?h.white there. Therefore, where there is insufficient GM/WM contrast, the wm.mgz cannot help out. I think that there is no change in the behavior between v5.3 and v6.0 in this aspect. I have seen many cases where wm.mgz edit does not help to improve ?h.white surfaces in the data processed by v5.3.
There is an -overlay option in mris_make_surfaces which explicitly modifies voxel values in input image for mris_make_surfaces (brain.finalsurfs.mgz as default in recon-all). However, this option only works for cases when wm.mgz is extended, not deleted: In voxels which have value 255 in wm.mgz, it replaces current voxel value in input image by desired value of white matter. This option is currently not used in recon-all. I would opt for implementation of similar option for deletion of voxels in wm.mgz: Where wm.mgz voxel values equal 1, to replace input image voxel values by desired values of GRAY matter. This should force ?h.white to directly follow edits of wm.mgz
As for editing of 001.mgz, beware the pitfall with editing images with mutually different geometry in FreeView I posted today:
http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg52729.html
Antonin Skoch
Dear All,
Regarding wm.mgz edits not incorporated into wm surface, which seems to have been reported on by several of us with FS 6.00. I followed Antonin's idea of editing 001.mgz by switching wm-intensity voxels clearly not wm (mostly in the subtemporal regions, at times close to bright dural portions in the crown) to gm intensities and ran recon-all -all, and this time all was OK. So it seems that the problem is that on re-initializing during -autorecon2-wm rerun, the routine does not edit out high intensity voxels even if manually excluded in wm.mgz. This seems to be different to some extent with how FS 5.3 was reacting, although I did not look at this systematically. There may be a reason that manual editing is just part of the story during rerun and does not automatically override the recon stream, but I do not know it. Octavian
---------- Forwarded message ---------- From: Octavian Lie octavian....@gmail.com Date: Sun, Apr 2, 2017 at 7:59 PM Subject: wm edits not incorporated To: octavian lie octavian....@gmail.com
Dear Bruce,
I just transferred subject r02.tar.gz.
Again, wm.mgz edits (all deleted voxels, no additions) are not incorporated on reruns using any of the recon-all -autorecon-wm -aitorecon3 recon-all -autorecon2 -autorecon3 recon-all -make all
FOr example, the complete command line for a rerun is
recon-all -autorecon2-wm -autorecon3 -3T -bigventricles -cw256 -subjid r02 -parallel > output.txt &
Here are some of the edited voxels which are not excluded from the wm on rerun:
-40.14, -2.67, 116.29 coronal slice 132 12.73, -23.67, 126.16 coronal slice 171 14.05, 12.33, 11.61 coronal slice 147
Thank you, this is very helpful. Octavian
On Thu, Apr 13, 2017 at 12:19 PM, Antonin Skoch ansk@ikem.cz wrote:
Dear Octavian,
the current philosophy with wm.mgz edits is not to directly affect voxel values in original T1 image.
The wm.mgz serves as starting point for the white surface estimation. The ?h.white surface is estimated according gradient in voxel values in brain.finalsurfs.mgz (which is, basically, preprocessed T1 image) using basically preprocessed surfaces obtained from wm.mgz (?h.orig) as a starting position where to search for intensity gradient. The editing of wm.mgz helps to find out the GM/WM itensity gradient and properly place ?h.white there. Therefore, where there is insufficient GM/WM contrast, the wm.mgz cannot help out. I think that there is no change in the behavior between v5.3 and v6.0 in this aspect. I have seen many cases where wm.mgz edit does not help to improve ?h.white surfaces in the data processed by v5.3.
There is an -overlay option in mris_make_surfaces which explicitly modifies voxel values in input image for mris_make_surfaces (brain.finalsurfs.mgz as default in recon-all). However, this option only works for cases when wm.mgz is extended, not deleted: In voxels which have value 255 in wm.mgz, it replaces current voxel value in input image by desired value of white matter. This option is currently not used in recon-all. I would opt for implementation of similar option for deletion of voxels in wm.mgz: Where wm.mgz voxel values equal 1, to replace input image voxel values by desired values of GRAY matter. This should force ?h.white to directly follow edits of wm.mgz
As for editing of 001.mgz, beware the pitfall with editing images with mutually different geometry in FreeView I posted today:
http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg52729.html
Antonin Skoch
Dear All,
Regarding wm.mgz edits not incorporated into wm surface, which seems to have been reported on by several of us with FS 6.00. I followed Antonin's idea of editing 001.mgz by switching wm-intensity voxels clearly not wm (mostly in the subtemporal regions, at times close to bright dural portions in the crown) to gm intensities and ran recon-all -all, and this time all was OK. So it seems that the problem is that on re-initializing during -autorecon2-wm rerun, the routine does not edit out high intensity voxels even if manually excluded in wm.mgz. This seems to be different to some extent with how FS 5.3 was reacting, although I did not look at this systematically. There may be a reason that manual editing is just part of the story during rerun and does not automatically override the recon stream, but I do not know it. Octavian
---------- Forwarded message ---------- From: Octavian Lie octavian....@gmail.com Date: Sun, Apr 2, 2017 at 7:59 PM Subject: wm edits not incorporated To: octavian lie octavian....@gmail.com
Dear Bruce,
I just transferred subject r02.tar.gz.
Again, wm.mgz edits (all deleted voxels, no additions) are not incorporated on reruns using any of the recon-all -autorecon-wm -aitorecon3 recon-all -autorecon2 -autorecon3 recon-all -make all
FOr example, the complete command line for a rerun is
recon-all -autorecon2-wm -autorecon3 -3T -bigventricles -cw256 -subjid r02 -parallel > output.txt &
Here are some of the edited voxels which are not excluded from the wm on rerun:
-40.14, -2.67, 116.29 coronal slice 132 12.73, -23.67, 126.16 coronal slice 171 14.05, 12.33, 11.61 coronal slice 147
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.
freesurfer@nmr.mgh.harvard.edu
-
Antonin Skoch -
Octavian Lie