On Oct 28, 2005, at 8:03 AM, Donna Hanlon wrote:
Hi all,
For non-human segmenters, .....
For the record, this actually refers to human segmenters of nonhuman
brains. We are hoping that Caret segmentation soon becomes
sufficiently automated that chimpanzees can handle it, but we're not
quite there yet....
DVE
yesterday I confirmed that if you segment from a volume whose voxdims
are not cubic 1mm, Caret generates a surface whose mm coordinates are
accurate (assuming the voxdims are accurate).
Example: With a cubic 0.5mm anatomical volume loaded in window 1 and
the resulting fiducial surface in window 2, click in volume window on
voxel along medial surface; then click on voxel 57 slices laterally
(keeping coronal and axial slices constant). Fiducial distance is
28.3 (roughly half of 57):
VOXEL IJK(1, 86, 41) XYZ(0.0, -5.0, 0.0)
View Window 2 Fiducial XYZ: -0.1, -5.3, 0.8
VOXEL IJK(58, 86, 41) XYZ(28.5, -5.0, 0.0)
View Window 2 Fiducial XYZ: 28.2, -5.5, 0.2 Distance: 28.3
Most of you probably took this for granted, but in the old SureFit
days, this wasn't true (unless you explicitly scaled the surface in
Caret). It has probably been true in Caret since segmentation was
added, but Caret used to prevent you from segmenting if voxdims were
not cubic 1mm. John made a change a while back to enable non-human
segmenters to avoid lying to Caret.
Donna
_______________________________________________
caret-users mailing list
caret-users@brainvis.wustl.edu
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
