Have you tried editing the header and applying to the image to fix the -1?
-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Robert C. Welsh, PhD
Research Investigator
Department of Radiology
University of Michigan
(734) - 764 - 2412 (fax)
[log in to unmask]
>>> Joseph Devlin <[log in to unmask]> 08/30/01 08:33AM >>>
Hi all,
I'm having some difficulty normalising certain structural scans and was
hoping someone might be able to help. They are 1x1x1.5mm T1 scans from
healthy subjects. Using DISPLAY they appear just fine on the screen with
the dimensions (256x256x128), data type (int16), and intensity (y=1x) all
ok. The voxel size is coming up at -1 x 1 x 1.5 which is odd but the
origin looks correct. When I try to normalise the image, it goes wrong
from the very beginning with the affine registration convergence values
around 30-50. In contrast, images which normalise correctly seem to start
around 90-100 and stay in that general region. When it begins the
nonlinear phase, the variance values in the 12 iterations go from 10,000 to
about 500. In contrast, when normalization works for my the variance
values are often less than 20. The end result is a brain shifted almost
out of the field of view.
I've tried reorienting the images by choosing to resize the x coord (in
DISPLAY) to -1 and then pressing the reorient button. Afterwards,
everything in DISPLAY looks correct but there is no difference to the
normalisation. I've normalised many brains before and not run into this
problem. Today I've seen it with 5 of my subjects! I'd appreciate any
suggestions as I'm running out of ideas.
Thanks in advance.
Joe
Joseph Devlin, Ph. D.
Department of Experimental Psychology
South Parks Road
Oxford University
Oxford OX1 3UD, U.K.
Phone: 01865 271 309
Email: [log in to unmask]
|