Standard best practices for creating binary GM, WM and CSF masks from freesurfer recon-all output
External Email - Use Caution
Hi there
I am new to Freesurfer and I would like some help with best practices for exporting binary GM, WM and CSF masks.
I have run recon-all (FS v7.1) on some T1w images. I would now like to create binary GM, WM and CSF masks from the recon-all results. I have an existing adapted script that uses AFNI’s *3dcalc* to match certain label values in *aparc+aseg.mgz* and output binary masks (I have since realized I can also use *mri_binarize* for that), see below for the exact label values that are matched and added together per tissue segmentation. However, I have also seen the output files such as *brainmask.mgz*, *ribbon.mgz*, *wm.mgz*; so I am uncertain about what the standard practices are for creating binary masks for GM, WM and CSF from freesurfer’s output.
I’ve searched a bit on the mailing list and found two similar questions: https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg46545.html and https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg11365.html. However, these suggest different approaches (e.g. inverting WM and VCSF to get GM, vs matching several structures and adding them together), and I am unsure whether these will end up covering the same voxels.
*So, my main question is:* is there a standard best practice for generating binary masks for GM, WM and CSF, i.e. which labels should be included or excluded? Or which standard outputs of freesurfer should be used?
Additionally I would also like to get a binary mask of the amygdala and the Brodmann areas 4a, 4p and 6. The amygdala seems to work fine. For the motor areas, I have tried my standard route of matching the label values (404, 405, 406 respectively) in aparc+aseg.mgz, but this doesn’t seem to yield any matches, i.e. all voxels in the resulting mask image have value zero. *My second question:* can you perhaps think of a reason for this error? *And question 2, part 2:* I assume BA 4a, 4p and 6 are all bilateral, since I didn’t see them being specified as LH or RH, is there a quick way to get only the LH?
Best regards,
Stephan Heunis
--------------------
GM, WM and CSF labels used in current script:
# Grey matter: # Subcortical nuclei: # - 10: Left-Thalamus-Proper # - 11: Left-Caudate # - 12: Left-Putamen # - 49: Right-Thalamus-Proper # - 50: Right-Caudate # - 51: Right-Putamen # Cerebellum: # - 8: Left-Cerebellum-Cortex # - 47: Right-Cerebellum-Cortex # Hippocampus and amygdala # - 17: Left-Hippocampus # - 18: Left-Amygdala # - 53: Right-Hippocampus # - 54: Right-Amygdala # Cortical ribbon: # - within(a,1000,3000): ctx-lh-[all], ctx-rh-[all]
# White Matter: # - 2: Left-Cerebral-White-Matter # - 7: Left-Cerebellum-White-Matter # - 41: Right-Cerebral-White-Matter # - 46: Right-Cerebellum-White-Matter # - 251: CC_Posterior # - 252: CC_Mid_Posterior # - 253: CC_Central # - 254: CC_Mid_Anterior # - 255: CC_Anterior
# CSF: # - 4: Left-Lateral-Ventricle # - 14: 3rd-Ventricle # - 43: Right-Lateral-Ventricle
On 9/14/2020 3:07 PM, Stephan Heunis wrote:
External Email - Use Caution
Hi there
I am new to Freesurfer and I would like some help with best practices for exporting binary GM, WM and CSF masks.
I have run recon-all (FS v7.1) on some T1w images. I would now like to create binary GM, WM and CSF masks from the recon-all results. I have an existing adapted script that uses AFNI’s/3dcalc/to match certain label values in/aparc+aseg.mgz/and output binary masks (I have since realized I can also use/mri_binarize/for that), see below for the exact label values that are matched and added together per tissue segmentation. However,I have also seen the output files such as/brainmask.mgz/,/ribbon.mgz/,/wm.mgz/; soI am uncertainabout what the standard practices are for creating binary masks for GM, WM and CSF from freesurfer’s output.
I’ve searched a bit on the mailing list and found two similar questions:https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg46545.htmland.... However, these suggest different approaches (e.g. inverting WM and VCSF to get GM, vs matching several structures and adding them together), and I am unsure whether these will end up covering the same voxels.
*So, my main question is:*is there a standard best practice for generating binary masks for GM, WM and CSF, i.e. which labels should be included or excluded? Or which standard outputs of freesurfer should be used?
You can use aparc+aseg.mgz with either 3dcalc or mri_binarize. If you are just using whole cortex, then you can use aseg.mgz instead (just saves you some typing since there are not a bunch of cortical parcellations).
Additionally I would also like to get a binary mask of the amygdala and the Brodmann areas 4a, 4p and 6. The amygdala seems to work fine. For the motor areas, I have tried my standard route of matching the label values (404, 405, 406 respectively) in aparc+aseg.mgz, but this doesn’t seem to yield any matches, i.e. all voxels in the resulting mask image have value zero.*My second question:*can you perhaps think of a reason for this error?*And question 2, part 2:*I assume BA 4a, 4p and 6 are all bilateral, since I didn’t see them being specified as LH or RH, is there a quick way to get only the LH?
We do have surface parcellations for 4a, 4p, and 6 (look in the labels folder), but they are not sampled into the volume. You can use mri_label2vol to create masks from them. Run it with --help to get examples
Best regards,
Stephan Heunis
GM, WM and CSF labels used in current script:
# Grey matter: # Subcortical nuclei: # - 10: Left-Thalamus-Proper # - 11: Left-Caudate # - 12: Left-Putamen # - 49: Right-Thalamus-Proper # - 50: Right-Caudate # - 51: Right-Putamen # Cerebellum: # - 8: Left-Cerebellum-Cortex # - 47: Right-Cerebellum-Cortex # Hippocampus and amygdala # - 17: Left-Hippocampus # - 18: Left-Amygdala # - 53: Right-Hippocampus # - 54: Right-Amygdala # Cortical ribbon: # - within(a,1000,3000): ctx-lh-[all], ctx-rh-[all]
# White Matter: # - 2: Left-Cerebral-White-Matter # - 7: Left-Cerebellum-White-Matter # - 41: Right-Cerebral-White-Matter # - 46: Right-Cerebellum-White-Matter # - 251: CC_Posterior # - 252: CC_Mid_Posterior # - 253: CC_Central # - 254: CC_Mid_Anterior # - 255: CC_Anterior
# CSF: # - 4: Left-Lateral-Ventricle # - 14: 3rd-Ventricle # - 43: Right-Lateral-Ventricle
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
External Email - Use Caution
Thanks, I'll get on it!
On Tue, Sep 15, 2020 at 4:16 PM Douglas N. Greve dgreve@mgh.harvard.edu wrote:
On 9/14/2020 3:07 PM, Stephan Heunis wrote:
External Email - Use CautionHi there
I am new to Freesurfer and I would like some help with best practices for exporting binary GM, WM and CSF masks.
I have run recon-all (FS v7.1) on some T1w images. I would now like to create binary GM, WM and CSF masks from the recon-all results. I have an existing adapted script that uses AFNI’s *3dcalc* to match certain label values in *aparc+aseg.mgz* and output binary masks (I have since realized I can also use *mri_binarize* for that), see below for the exact label values that are matched and added together per tissue segmentation. However, I have also seen the output files such as *brainmask.mgz*, *ribbon.mgz*, *wm.mgz*; so I am uncertain about what the standard practices are for creating binary masks for GM, WM and CSF from freesurfer’s output.
I’ve searched a bit on the mailing list and found two similar questions: https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg46545.html and https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg11365.html. However, these suggest different approaches (e.g. inverting WM and VCSF to get GM, vs matching several structures and adding them together), and I am unsure whether these will end up covering the same voxels.
*So, my main question is:* is there a standard best practice for generating binary masks for GM, WM and CSF, i.e. which labels should be included or excluded? Or which standard outputs of freesurfer should be used?
You can use aparc+aseg.mgz with either 3dcalc or mri_binarize. If you are just using whole cortex, then you can use aseg.mgz instead (just saves you some typing since there are not a bunch of cortical parcellations).
Additionally I would also like to get a binary mask of the amygdala and the Brodmann areas 4a, 4p and 6. The amygdala seems to work fine. For the motor areas, I have tried my standard route of matching the label values (404, 405, 406 respectively) in aparc+aseg.mgz, but this doesn’t seem to yield any matches, i.e. all voxels in the resulting mask image have value zero. *My second question:* can you perhaps think of a reason for this error? *And question 2, part 2:* I assume BA 4a, 4p and 6 are all bilateral, since I didn’t see them being specified as LH or RH, is there a quick way to get only the LH?
We do have surface parcellations for 4a, 4p, and 6 (look in the labels folder), but they are not sampled into the volume. You can use mri_label2vol to create masks from them. Run it with --help to get examples
Best regards,
Stephan Heunis
GM, WM and CSF labels used in current script:
# Grey matter: # Subcortical nuclei: # - 10: Left-Thalamus-Proper # - 11: Left-Caudate # - 12: Left-Putamen # - 49: Right-Thalamus-Proper # - 50: Right-Caudate # - 51: Right-Putamen # Cerebellum: # - 8: Left-Cerebellum-Cortex # - 47: Right-Cerebellum-Cortex # Hippocampus and amygdala # - 17: Left-Hippocampus # - 18: Left-Amygdala # - 53: Right-Hippocampus # - 54: Right-Amygdala # Cortical ribbon: # - within(a,1000,3000): ctx-lh-[all], ctx-rh-[all]
# White Matter: # - 2: Left-Cerebral-White-Matter # - 7: Left-Cerebellum-White-Matter # - 41: Right-Cerebral-White-Matter # - 46: Right-Cerebellum-White-Matter # - 251: CC_Posterior # - 252: CC_Mid_Posterior # - 253: CC_Central # - 254: CC_Mid_Anterior # - 255: CC_Anterior
# CSF: # - 4: Left-Lateral-Ventricle # - 14: 3rd-Ventricle # - 43: Right-Lateral-Ventricle
Freesurfer mailing listFreesurfer@nmr.mgh.harvard.eduhttps://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
-
Douglas N. Greve -
Stephan Heunis