You can use FNIRT to register the FA to the T1 and then apply the warpfield
generated to unwarp the raw data:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;XnD16g;20080923081452%2B0
100
To do this right, you may want to register your FA to T1 with FLIRT (I would
use 6 dof for this), then register the output FA to the T1 with FNIRT,
invert the FLIRT transform and then transform the warpfield into the
original diffusion space using the inverted FLIRT transform, so that you can
avoid needing to rotate your gradient table, and then unwarp the raw data.
Alternatively, you will need to apply the rotational components of the
original FLIRT linear transform to your bvecs file which can be found using
avscale --allparams and there is a rotate_bvecs script that you can use to
do the rotation.
This process is fairly complex, and definitely not ideal. It is much better
to acquire field maps or correct for distortion by acquiring phase up and
phase down averages. A lot of papers do not correct for EPI distortion, so
one can also choose not to do this. If you don't correct for it, though,
you cannot rely on good alignment between your diffusion and structural data
in particular brain regions, like the temporal lobe and the orbitofrontal
region. Depending on what you are trying to do with the data, this can be a
problem or not.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Eunjin Yoon
Sent: Monday, April 19, 2010 3:49 AM
To: [log in to unmask]
Subject: [FSL] EPI distortion of DTI
Hello, all
I'm beginner of FSL and DTI processing.
I am really wondering how to process distorted diffusion tensor images.
My images from 3T phillips scanner show some distortion in frontal area.
I want to correct this distortion, but I don't have filed maps.
And most of papers using DTI, they didn't mention about how they dealt with
this type of problem.
First of my problem during DTI processing is making brain mask using BET.
Is it necessary that all of the distorted part is included as brain?
If some part of the distorted areas cut off from the brain, is that problem
for following process or if all of the distorted areas are included as
brain, will it make a problem?
OR Should I correct the distortion?
If I should do that, how to do that when I don't have field maps?
Thanks in advance.
EJ
|