Hi Bruce,
Yes there is a major inhomogeneity issue that we deal with prior to running FS, along with an addition skull strip step. With all of our 4T data, because of the large bias field and hyper-intense Dura Matter, we perform a bias correction and skull-strip before the data is entered into FS. During these pre-processing steps we think that some orientation information is lost due to multiple format conversions. Forgive me but I am introducing other issues, but because of this loss of positional information I cannot reliably optimize the talairach transformation in tkregister2 (see attached picture). You will see that the reference point in the FOV of tkregister2 is in the wrong quadrant. Though from what I can see the talairach transforms seem to be good, but some appear to be slightly underestimated. So maybe this is probably a good target to try and fix the underestimation of our surfaces. This problem with viewing the talairach has never sat well with me, but the surfaces and segmentations in our other studies look good despite this issue. Is there a way to reset the FOV in tkregister so that I can reliably correct any slight errors in the talairach transform to see if this can fix what were seeing with this subject group?
When I add control points in the area where wm should be, the segmentation that usually results includes anterior carotid artery or lots of the small pieces of dura that I was unable to threshold out. Our next option is to simply edit all non-brain tissue around these areas of atrophy, but if there is any other relatively less time consuming option then it would be much appreciated.
Sky
At 12:52 PM 7/16/2008, Bruce Fischl wrote:
Hi Sky,
hard to tell from just the one slice. Is there a big bias field in the data? Have you tried adding control points in the wm that you lose?
cheers, Bruce
On Wed, 16 Jul 2008, Sky Raptentsetsang wrote:
Hello Surfers,
I am trying to FreeSurfer process a group of subjects from our Frontal Temporal Lobar Dementia project that were scanned on our 4T Siemens MRI. So far this endeavor has only created long hours of edits due to the absence of any visible White Matter in the Frontal and/or Temporal Poles. I have included a couple pictures of FS surface estimate on two of our many problematic subjects. I am wondering if you have any solutions or input to help FS include these regions of great atrophy.
Thanks in advance for any input on this tedious issue.
Sky R.
