You may need to flip (negate) one of the x, y, or z axes in the gradient
table. After running DTIFIT, you can then load the FA and 2 V1 images in
FSLView setting one to lines and the other to colors modulated by FA
intensity. Then check to see that the lines flow smoothly and continuously
in the corpus callosum in the coronal and axial planes, and the corona
radiata in the sagittal plane. If you need to negate (or flip) one of the
gradient directions, the diffusion lines will be 180 opposite of what they
should be in 2 of the 3 planes. The plane that looks correct will not
contain the dimension that needs to be flipped (e.g. the axial plane
contains x and y dimensions, and not z, so if axial is correct and coronal
and sagittal are not, then z would need to be negated).
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Jaymin Upadhyay
Sent: Thursday, September 13, 2007 1:57 PM
To: [log in to unmask]
Subject: [FSL] Coronal DTI
Hi,
I recently acquired and analyzed coronal DTI data using a 72 direction DTI
pulse sequence on a
Siemens scanner. After analyzing the data I saw that the color coded FA
maps looked fine; mid-
sagittal corpus callosum section were depicted in red, corticospinal tract
in blue, longitudinal
fasciculus in green, etc. When I looked at the principle eigenvectors I saw
that they were not correct
and appeared quite different in comparison to DTI data obtain axially. To
check if the eigenvectors
did not simply appear different because of very small differences in voxel
dimensions between axial
and coronal DTI dataset, I performed probabilistic mapping of the corpus
callosum and got wierd
results. To analyze my coronal data I didn't reslice or reorient the
volumes and I used the bvec and
bval files extracted from the coronal DTI dataset and not the axial dataset.
Is there an explanation of
why the color coded FA maps would look fine and at the same time V1 maps are
incorrect. I can
send images if needed.
Thanks,
Jaymin
|