Hi Bruce, Can you elaborate briefly on the 6-connected bit? e.g., if one places a CP manually in a voxel, and that CP is not face-connected on all sides to either other manually-placed CPs, or CPs determined automatically by FS, then it doesn't get counted as a CP after all?
Is there any volume that allows one to see the automatically generated CPs?
Something definitely seems wonky about the inclusion of the manually- specified CPs as part of the aseg normalization (mri_ca_normalize), because the resulting norm.mgz (and thus eventually brain.mgz) ends up differing appreciably in locations distant from the manually placed CPs. In contrast, if you skip the re-creation of the norm.mgz by using the - nocanorm flag, the resulting brain.mgz is much more reasonably behaved, with its largest differences in the vicinity of the CPs.
thanks, -MH
On Fri, 2012-01-27 at 14:58 -0500, Bruce Fischl wrote:
Hi Mike
it's a bit hard to state the region of effect for the control points. Essentially we go through and label voxels as control points or not based on their intensity, intensity gradient and connectivity (that is, the must be 6-connected to other control points) then build a Voronoi diagram and each control point sets the scaling for its Voronoi triangle. Thus if you have a control point surrounded by others its region of effect is small, but one control point all by itself can have a large region of effect.
Also, 5.1 applies the manually specified control points to the aseg normalization (norm.mgz), whereas older versions didn't. Not everyone is happy with this, so I think there is a backwards compatibility flag. Nick would know.
Bruce
On Fri, 27 Jan 2012, Sabin Khadka wrote:
Hi Michael,I had the same problem too. It might be because of the type of scanner you are using. I added -washu_mprage flag, it pretty much helped me (I did not had to add a lot of controls points and so on so forth. You can go through the link below.
https://mail.nmr.mgh.harvard.edu/pipermail//freesurfer/2009-August/011695.ht...
Hope it helps.
-SK
On Fri, Jan 27, 2012 at 12:06 PM, Michael Harms mharms@conte.wustl.edu wrote:
Hi guys, We are currently trying to fix some errors in the white/pial surfaces where there are thin white matter strands by using control points, and are noticing a couple things: 1) The resulting WM surface in the area of the CPs can end up too far into the GM instead. Given that, is there any practical guidance for how to think about the surrounding spatial extent that is impacted by a given CP? i.e., How do CP's actually get used within mri_normalize in an algorithmic sense? 2) The surfaces are being impacted in places distant from the CPs. e.g., CP's placed in the left anterior temporal lobe are resulting in surface changes in the right anterior temporal lobe. And when I difference the original norm.mgz vs. the one obtained after using CP's, I'm seeing an odd pattern of intensity differences which is clearly not limited to just the area of the CP's (which would be my expectation). This is version 5.1. thanks, -MH -- Michael Harms, Ph.D. -------------------------------------------------------------------- Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 Renard Hospital, Room 6604 Tel: 314-747-6173 660 South Euclid Ave. Fax: 314-747-2182 St. Louis, MO 63110 Email: mharms@wustl.edu -------------------------------------------------------------------- _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.