Nick, Doug or anyone with thoughts on the matter,
This is a return to a similar question I raised in June-2nd thread "Some FS input questions", but a little less awkward.
We'd like to present to FS already brain-extracted volumes that are already "conformed", ie: 256^3, and 1 mm isotropic
This would appear to avoid the necessity to turn off aseg (mentioned by Nick in previous thread).
I also see the enticing looking options like "noskullstrip", but... I'm wary of two things:
a) In general, when you select a "nosomething" option, does recon-all do the right thing to "pass the data through" (usually a copy I suspect) so that subsequent steps receive their input file(s)?
b) There may be steps that expect the skull or neck to be in place? "EM Registration with Skull" (skull-lta) sounds like it might be one? So I'd like to know if there are such, and whether I should care?
According to my chart here: grahamwideman.com/gw/brain/fs/fsunderstanding2006/processvsdata2006.htm ...looks like I *should* be able to bypass rmneck and skull-lta (or just let them run unsatisfactorily) with no downstream impact.
Any thoughts?
Thanks,
Graham
Hi Graham,
the EM Registration with Skull is for generating ICV measures, which of course you won't be able to get with the skull gone.
Bruce On Thu, 13 Jul 2006, Graham Wideman wrote:
Nick, Doug or anyone with thoughts on the matter,
This is a return to a similar question I raised in June-2nd thread "Some FS input questions", but a little less awkward.
We'd like to present to FS already brain-extracted volumes that are already "conformed", ie: 256^3, and 1 mm isotropic
This would appear to avoid the necessity to turn off aseg (mentioned by Nick in previous thread).
I also see the enticing looking options like "noskullstrip", but... I'm wary of two things:
a) In general, when you select a "nosomething" option, does recon-all do the right thing to "pass the data through" (usually a copy I suspect) so that subsequent steps receive their input file(s)?
b) There may be steps that expect the skull or neck to be in place? "EM Registration with Skull" (skull-lta) sounds like it might be one? So I'd like to know if there are such, and whether I should care?
According to my chart here: grahamwideman.com/gw/brain/fs/fsunderstanding2006/processvsdata2006.htm ...looks like I *should* be able to bypass rmneck and skull-lta (or just let them run unsatisfactorily) with no downstream impact.
Any thoughts?
Thanks,
Graham _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Graham,
In general, you should be careful when selecting the '-nosomething' options, as recon-all does not perform much checking for these flags (only the -autorecon1,2,3 flags contain the validity checks).
An option when experimenting with the flags is to use the -dontrun flag, which just prints all the commands it will run without executing. Then you can manually check for validity.
Also, perusing the recon-all script is the next best thing. In the case of the -noskullstrip flag, this just skips the mri_watershed block of code, so that means brainmask.auto.mgz is not created.
This means, I suppose, that you could just name your own skullstripped volume brainmask.auto.mgz (and copy that to brainmask.mgz, which is the one that should be edited, if need be), and then run:
recon-all -autorecon2 -normneck -noskull-lta
Nick
On Thu, 2006-07-13 at 13:10 -0700, Graham Wideman wrote:
Nick, Doug or anyone with thoughts on the matter,
This is a return to a similar question I raised in June-2nd thread "Some FS input questions", but a little less awkward.
We'd like to present to FS already brain-extracted volumes that are already "conformed", ie: 256^3, and 1 mm isotropic
This would appear to avoid the necessity to turn off aseg (mentioned by Nick in previous thread).
I also see the enticing looking options like "noskullstrip", but... I'm wary of two things:
a) In general, when you select a "nosomething" option, does recon-all do the right thing to "pass the data through" (usually a copy I suspect) so that subsequent steps receive their input file(s)?
b) There may be steps that expect the skull or neck to be in place? "EM Registration with Skull" (skull-lta) sounds like it might be one? So I'd like to know if there are such, and whether I should care?
According to my chart here: grahamwideman.com/gw/brain/fs/fsunderstanding2006/processvsdata2006.htm ...looks like I *should* be able to bypass rmneck and skull-lta (or just let them run unsatisfactorily) with no downstream impact.
Any thoughts?
Thanks,
Graham _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
Graham Wideman -
Nick Schmansky