I am not clear on what you are trying to compare here, tracts or ROIs used
to generate tracts? If you want volume measurements, why not measure in
individual subject space. If you want to see tract overlap across subjects,
why not use as many degrees of freedom as possible to maximize your ability
to align the tracts? E.g., if your individual results are in diffusion
space, you could use the initial processing steps of TBSS to generate
warpfields between each subject's diffusion space and a standard space.
Even if you do this, there will be significant individual variation, but at
least the tracts will be as well aligned as is currently possible.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Dylan Archbold Alegria
Sent: Thursday, May 07, 2009 5:57 PM
To: [log in to unmask]
Subject: [FSL] Registration for Fiber Tracking AnalysisRegistration for
Fiber Tracking Analysis
Hi,
I'm doing my first analysis using using probtrackx. I was hoping to compare
the ROIs generated from tracking in a common space. I've thresholded them
in a way that accounts for variation in "trackability." I'm trying to
determine the appropriate DOF for my registration to compare overlap.
I'm wondering why the standard DOF for registration is 12. Doesn't that
drastically change the size and shape of your ROIs, effectively diminishing
individual variation?
To compare the size/overlap of tracts I think that I would do a 7 or 9 DOF
to control for total brain size, but not distort the tract ROI too much. Is
my logic right or are there good reasons to do a 12 DOF registration?
Thanks for the help!
Dylan
Dylan Alegria
CIBSR Imaging RA
Stanford University
[log in to unmask]
(650) 736 0732
|