Hi Jenni--as it turns out, FYI, a couple hundred control points helps on many of these subjects, just not all. The aseg's seem to be for the most part just fine, so I won't mess with it.
Thanks, Jess
-----Original Message----- From: Jenni Pacheco [mailto:jpacheco@nmr.mgh.harvard.edu] Sent: Friday, June 08, 2007 4:51 AM To: Turner, Jessica Cc: freesurfer@nmr.mgh.harvard.edu Subject: RE: [Freesurfer] question re: low contrast images
Hi Jess,
I think there is a way to fix the aseg, you can use the control points with the mri_ca_normalize flag as part of the aseg. I've not used the command much, but you should be able to add the flag '-f tmp/control.dat' to the normal mri_ca_normalize command to run this step. We are working
on adding this to recon-all, but currently it's not there. You would have to run mri_ca_normalize seperately.
Jenni
On Thu, 7 Jun 2007, Turner, Jessica wrote:
Hi Jenni--Ok, I'm running a couple with control points and I'll see
how
they come out. Yes, the aseg is not any better--do I need to fix that separately or will the control points work on that??
Thanks, Jess
-----Original Message----- From: Jenni Pacheco [mailto:jpacheco@nmr.mgh.harvard.edu] Sent: Thursday, June 07, 2007 11:00 AM To: Turner, Jessica Cc: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] question re: low contrast images
Hi Jess,
It certainly is a problem, from the snapshots though it looks like
there
is some visible contrast there, so control points should help. be
sure
to only place them in true WM regions, and avoid partial volumed voxels.
A
few spread out on each slice should do the trick - no need to overload them. Then you can run recon-all -autorecon2-cp, and see if it fixes your surfaces.
How does the aseg look? I'd imagine it's pretty bad as well.
Jenni
On Thu, 7 Jun 2007, Turner, Jessica wrote:
Hi experts-I have some very low-contrast images from an older dataset and the Freesurfer segmentation completely misses the lower and posterior bits of the occipital lobe on many of the subjects (see attached images from an example subject). What is the recommendation
in
this case? Can I fix this with control points? It's a much more extensive failure than I'm used to seeing fixed with control points.
Thanks,
Jess
Jessica Turner, Ph.D.
Project Scientist, FBIRN (www.nbirn.net)
Department of Psychiatry and Human Behavior
University of California, Irvine
Phone: (949) 824-3331
Fax: (949) 824-3324