Hi Eric,
Based on what you describe, you probably have some non-uniformity in
your volume that bias correction may help. I use FSL's fast for this
purpose (which uses BET as preprocessing), but there are many other
options (e.g., MNI's nu_correct and AFNI's 3dUniformize). You can
upload your volume here and I can either confirm or rule out this issue:
http://pulvinar.wustl.edu/cgi-bin/upload.cgi
Peak tweaking makes a huge difference in segmentation quality. This
page might be helpful:
http://brainvis.wustl.edu/help/peak_tweaking/
If you're getting good results out of Freesurfer, then you can use the
resulting surface in Caret (e.g., for registration, visualization, or
other purposes). Except in cases where the quality of the structural
MRI is exceptionally poor (e.g., acquisition resolution way to low;
non-uniformity beyond hope -- like the surface coil scans I used to
see), then bias correction and peak tweaking usually give good results
in Caret.
On 03/31/2006 11:03 AM, Faden, Eric (NIH/NIMH) [F] wrote:
I have tried two different approaches thus far and neither seems to yield
reasonable results. I first tried to feed in the full scan in ACPC coords.
The image was rotated in brains, converted to AFNI, and then imported to caret.
It is correctly oriented and looks ok although the contrast seems a bit off
(some of the image seems to white). After cutting the brain into just the left
hemi I ran it through the segmentation process which yielded a pretty bad
surface that still included about 90% of the eyes and a very large chunk of the
skull. I then decided to try it with just a brain image (as I already had a
brain mask in brains). Following the same steps as the first attempt with the
brain only image also yielded a fairly bad initial surface as well as leaving a
large block of the hindbrain. Are there any other tricks people could provide?
Is there a trick to picking the gray/white matter peaks? For reference I have
run the same image through a modified freesurfer surface generation which works
fairly well, but was curious to see if caret could top it.
-Eric
------------------------------------------------------------------------
_______________________________________________
caret-users mailing list
caret-users@brainvis.wustl.edu
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
--
Donna L. Dierker
(Formerly Donna Hanlon; no change in marital status -- see
http://home.att.net/~donna.hanlon for details.)
