That looks fine. You should be able to just use one of the bvecs files so
long as they are basically the same. The bvals is the same for both.
There has been recent debate as to whether you should correct bvecs files
for angular rotations introduced by eddy_correct, and I posted some code
capable of doing this yesterday (though I am not convinced that it matters).
You could concatenate your bvecs files into a file with 218 directions, use
the eddy_correct script that I just posted and correct both the data and
bvecs at the same time. Finally, you would skip the fslroi and fslmaths
steps below.
Some would say that that is a more correct way to do things, but your
processing time will be longer for bedpostX and it is unclear if it will
make any difference anyways.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Rajeet Saluja
Sent: Monday, March 30, 2009 12:51 PM
To: [log in to unmask]
Subject: [FSL] averaging DTI acquisitions
Hi,
I wanted to average 2 DTI acquisitions and was a little confused about a
couple of things. After going through the archives I found a method of
performing the averaging procedure (as posted by Matt), but had a few
questions about it. If I take 2 files called data1.nii.gz and data2.nii.gz
(converted from dicom using dicom2nii from mricron) and I have 109 volumes
(99 directions and 10 b0), I should perform the following steps:
fslmerge -t data_av.nii.gz data1.nii.gz data2.nii.gz
eddy_correct data_av.nii.gz data_av_corr.nii.gz 0
fslroi data_av_corr.nii.gz av1_corr.nii.gz 0 109
fslroi data_av_corr.nii.gz av2_corr.nii.gz 109 109
fslmaths av1_corr.nii.gz -add av2_corr.nii.gz -div 2 data_av_final.nii.gz
Is this correct?
If it is correct, what do I do with the two sets of bvals and bvecs files
that
were created by dicom2nii? Do I just use one of them for the further
analysis
or do I have to perform other manipulations of these files?
Also, if I were to perform eddy_correct on each of the data sets then
combined them with the last line of the above procedure, wouldn't that
essentially do the same thing? Is there any benefit of merging, correcting,
splitting, and then averaging vs. correcting each separately then just
averaging at the end?
Thank you in advance for your help.
Rajeet
|