Running eddy_correct on a big, merged, 4D volume (what I am guessing you saw
that I suggested previously) accomplishes the same thing as eddy_correcting
the individual datasets before averaging. The main reason to average before
eddy correct (however after rigid body motion correction) is if your SNR is
really low and eddy_correct is prone to failing.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Kochunov, Peter
Sent: Saturday, September 20, 2008 7:22 PM
To: [log in to unmask]
Subject: Re: [FSL] eddy_correct before are after averaging
Your physicist is giving you a sound advice. The most conservative way would
be to process each set separately and merge them at the point of final
results e.g. to average the FA images. However, this won't work if you're
planning to do fiber tracking.
pk
________________________________
From: FSL - FMRIB's Software Library on behalf of Michael Scheel
Sent: Sat 9/20/2008 6:48 PM
To: [log in to unmask]
Subject: Re: [FSL] eddy_correct before are after averaging
Hi Markus, thanks for the answer. The reason I am asking was that my
local physists told me it would be wise to do the eddy correction for
every run seperately since the eddy currents could be very different
from one run to the other.
I used the suggestion that Matt gave in the forum archives and found
them very convenient. Now my physists got me puzzled (as this is
sometimes the case).
So my question probably really is: Do the eddy currents change from run
to run? Then the advice of running eddy current seperately would make sense.
Thanks, Michael
Markus Gschwind schrieb:
> Hi!
>
> Concatenate them all together with fslmerge -t, then eddy_correct the
> whole 4D file and then average them, after the eddy_correct.
>
> Hth, (there are some posts about this in the list, from e.g. Matt Glasser)
> Markus
>
>
> 2008/9/21 Michael Scheel <[log in to unmask]>:
>> Hi fsl-expert,
>>
>> we have acquired the same dti sequence 3 times for averaging. I'd like to
>> know if it is recommended to merge these 3 averages in one big file and
then
>> run eddy_correct OR first run eddy_correct on every dataset and then
average
>> them.
>>
>> Thanks, Michael
>>
>
>
>
|