I'm working on the latest developmental release for Rh. 9. and I'm having problems with the normalization procedures on my datasets. My datasets have large intensity inhomogeneities due to their acquisition with a surface coil (8-ch. GE). To fix this problem I have resorted to the use of control points. This works if I add enough points across the brain ( over 200), and I get a better normalization result after running "recon-all -normalization -usecontrolpoints" and another skull-stripping on the new T1-volume. To test whether or not this produces a good segmentaition of white matter, I run mri_segment on my brain.mgz -volume. If I'm satisfied with the segmentation result, I move on to the surface processing stage, and run the -autorecon2 -script. The problem is that after running the -autorecon2, the normalization is completely wrong again, looking more like the first normalization done without control points, and the wm.mgz -volume is also bad. So are the surfaces. How do I make freesurfer produce the same good results as I got during the first normalization? Do I need to specify the use of control points again, even though the brain.mgz volume looks ok?
Thanks,
Martin Ystad Medical Student University of Bergen Institute of Biomedicine Jonas Lies vei 91, 5009 Bergen, Norway.
Hi Martin,
we're looking into it - it's probably a bug in recon-all.
Bruce On Mon, 21 Nov 2005, Martin Ystad wrote:
I'm working on the latest developmental release for Rh. 9. and I'm having problems with the normalization procedures on my datasets. My datasets have large intensity inhomogeneities due to their acquisition with a surface coil (8-ch. GE). To fix this problem I have resorted to the use of control points. This works if I add enough points across the brain ( over 200), and I get a better normalization result after running "recon-all -normalization -usecontrolpoints" and another skull-stripping on the new T1-volume. To test whether or not this produces a good segmentaition of white matter, I run mri_segment on my brain.mgz -volume. If I'm satisfied with the segmentation result, I move on to the surface processing stage, and run the -autorecon2 -script. The problem is that after running the -autorecon2, the normalization is completely wrong again, looking more like the first normalization done without control points, and the wm.mgz -volume is also bad. So are the surfaces. How do I make freesurfer produce the same good results as I got during the first normalization? Do I need to specify the use of control points again, even though the brain.mgz volume looks ok?
Thanks,
Martin Ystad Medical Student University of Bergen Institute of Biomedicine Jonas Lies vei 91, 5009 Bergen, Norway.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi, I've encountered similar problems. Do you know if it's specific to the autorecon2 flag? I'm wondering if running each process separately and sequentially get around it, e.g, running recon-all -normalization2, then recon-all -segmentation, then recon-all -fill, etc...
Thanks, Alex
Alex Fornito M.Psych/PhD (clin. neuro.) candidate Melbourne Neuropsychiatry Centre and Department of Psychology The University of Melbourne alexander.fornito@wh.org.au
-----Original Message----- From: freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Bruce Fischl Sent: Mon 21/11/2005 11:03 PM To: Martin Ystad Cc: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Normalization problem
Hi Martin,
we're looking into it - it's probably a bug in recon-all.
Bruce On Mon, 21 Nov 2005, Martin Ystad wrote:
I'm working on the latest developmental release for Rh. 9. and I'm having problems with the normalization procedures on my datasets. My datasets have large intensity inhomogeneities due to their acquisition with a surface coil (8-ch. GE). To fix this problem I have resorted to the use of control points. This works if I add enough points across the brain ( over 200), and I get a better normalization result after running "recon-all -normalization -usecontrolpoints" and another skull-stripping on the new T1-volume. To test whether or not this produces a good segmentaition of white matter, I run mri_segment on my brain.mgz -volume. If I'm satisfied with the segmentation result, I move on to the surface processing stage, and run the -autorecon2 -script. The problem is that after running the -autorecon2, the normalization is completely wrong again, looking more like the first normalization done without control points, and the wm.mgz -volume is also bad. So are the surfaces. How do I make freesurfer produce the same good results as I got during the first normalization? Do I need to specify the use of control points again, even though the brain.mgz volume looks ok?
Thanks,
Martin Ystad Medical Student University of Bergen Institute of Biomedicine Jonas Lies vei 91, 5009 Bergen, Norway.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://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
-
Bruce Fischl -
Fornito, Alexander -
Martin Ystad