Folks:
I'm hoping that there's an aparc --> voxels program. That program might be mri_aparc2aseg, but I wanted to check first if that's the best way to go and whether there are particular options to know about and so on.
Basically I want a program that inputs a parcellation, and outputs a label mri where each voxel that's fully or partially between the white and pial surface gets labelled with the parcellation label from the vertex (white or pial) that it's closest to. (A variation that would also work for our purpose would label voxels even beyond that, possibly all voxels).
This seems like it's probably already been done, hence my asking.
I'm thinking that *might* be what mri_aparc2aseg does... but I'm not clear what the wiki description is trying to say:
https://surfer.nmr.mgh.harvard.edu/fswiki/mri_5faparc2aseg?action=highlight&...
First, it appears from Example 1 that mri_aparc2aseg does something interesting when run with basically no specific command line arguments, so presumably using default input files. But then what is the purpose of --ribbon?
Then the description talks about the automatic segmentation volume as though it's an input separate from the ?h.ribbon, yet there are no parameters related to that. And finally the --noribbon flag is mentioned as turning something off, but it's not clear what it's turning off.
Bottom line: I'd appreciate some clarification as to which program does what I want, and if it's mri_aparc2aseg, then what options are appropriate to do the simple version of the task where it ignores existing aseg and ribbon.
Thanks,
Graham
This is done automatically with recon-all. It creates aparc+aseg.mgz using mri_aparc2aseg. The ribbon is used to constrain the aparc in aparc+aseg to be in the ribbon defined by the surfaces instead of the ribbon defined by the aseg.
doug
Graham Wideman wrote:
Folks:
I'm hoping that there's an aparc --> voxels program. That program might be mri_aparc2aseg, but I wanted to check first if that's the best way to go and whether there are particular options to know about and so on.
Basically I want a program that inputs a parcellation, and outputs a label mri where each voxel that's fully or partially between the white and pial surface gets labelled with the parcellation label from the vertex (white or pial) that it's closest to. (A variation that would also work for our purpose would label voxels even beyond that, possibly all voxels).
This seems like it's probably already been done, hence my asking.
I'm thinking that *might* be what mri_aparc2aseg does... but I'm not clear what the wiki description is trying to say:
https://surfer.nmr.mgh.harvard.edu/fswiki/mri_5faparc2aseg?action=highlight&...
First, it appears from Example 1 that mri_aparc2aseg does something interesting when run with basically no specific command line arguments, so presumably using default input files. But then what is the purpose of --ribbon?
Then the description talks about the automatic segmentation volume as though it's an input separate from the ?h.ribbon, yet there are no parameters related to that. And finally the --noribbon flag is mentioned as turning something off, but it's not clear what it's turning off.
Bottom line: I'd appreciate some clarification as to which program does what I want, and if it's mri_aparc2aseg, then what options are appropriate to do the simple version of the task where it ignores existing aseg and ribbon.
Thanks,
Graham
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Doug:
Thanks for your quick reply -- but I still have some big holes in understanding, probably because I'm having a hard time digesting the mri_aparc2aseg's wiki page.
For the ribbon-related parameters we seem to have the 3 choices:
no ribbon param ... or ... --ribbon ...or... --noribbon
The Description presumably applies to using no ribbon parameter, and describes that *both* aseg *and* the surface-derived ?h.ribbon are used. If that's the case, then what does the --ribbon arg add?
It goes on to say "This can be turned off with --noribbon". What can be turned off? Use of ?h.ribbon? Or relabeling of the "unknown" voxels to zero (ie: aparc+aseg would have them labeled as "cortex").
Finally, for my task where I hope to ignore aseg entirely, is my solution to fake up an aseg volume filled with "cortex" voxels, and use that for input? (And I'm not seeing an argument to tell mri_aparc2aseg to use a different aseg...).
Thanks,
Graham
Graham Wideman wrote:
Doug:
Thanks for your quick reply -- but I still have some big holes in understanding, probably because I'm having a hard time digesting the mri_aparc2aseg's wiki page.
There's also the --help (might be the same thing).
For the ribbon-related parameters we seem to have the 3 choices:
no ribbon param ... or ... --ribbon ...or... --noribbon
Actually, if you do not spec a ribbon, then it is the same as --noribbon, which means that it uses the aseg-based ribbon.
The Description presumably applies to using no ribbon parameter, and describes that *both* aseg *and* the surface-derived ?h.ribbon are used. If that's the case, then what does the --ribbon arg add?
Turns off using the surface-based ribbon.
It goes on to say "This can be turned off with --noribbon". What can be turned off? Use of ?h.ribbon? Or relabeling of the "unknown" voxels to zero (ie: aparc+aseg would have them labeled as "cortex").
Finally, for my task where I hope to ignore aseg entirely, is my solution to fake up an aseg volume filled with "cortex" voxels, and use that for input? (And I'm not seeing an argument to tell mri_aparc2aseg to use a different aseg...).
Yes, that should work. You can use the ?h.ribbon files (change the values to 3 so aparc2aseg knows it's cortex).
doug
Thanks,
Graham
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Doug Greve -
Graham Wideman