If you compare the raw data before and after eddy_correct using the movie
tool in fslview, is the observed motion better or worse? Also, I don't know
what you mean by applying eddy_correct with and without motion correction.
Eddy_correct by design just aligns the images to the first b0, regardless of
if the misregistration is caused by eddy currents or subject motion.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Igor Yakushev
Sent: Thursday, July 02, 2009 7:21 AM
To: [log in to unmask]
Subject: [FSL] FDT: fewer fibers following motion correction - why?
Dear Colleagues,
I'm new to both FSL and fiber tracking. We are now testing different ways of
motion correction for our DTI data. We aim to perform tractography of the
cingulum bundle using deterministic algorithms (such as FACT). The data have
been acquired with 1.5 T Magnetom Sonata, Siemens in (just) 6 directions.
When we apply FDT' eddy current correction (ECC) without motion correction
(reference volume 0), we achieve an increase in the number and lengthy of
tracked fibers - fine! As expected, the mean FA values of the whole track
(cingulum) decrease and this decrease is accounted for by the higher number
and lengthy of fibers.
However, when we perform ECC with motion correction (with reference to B0
image), the number of subsequently tracked fibers decreases on average by
20%, the lengthy by ca. 10%. The mean FA of the track inscreases, but this
increase is not completely accounted for by the lower number and lengthy of
the fibers.
1) are these changes following motion correction to expect?
2) does motion correction as step in ECC implement a kind of smoothing? Why
do mean FA values decrease (by about 10%), although the number of
subsequebtly tracked fibers remains the same (or even lower) as when
tracking without any preprocessing (i.e. ECC, motion correction)?
Unfortunately, I couldn't find answers to these question in the archive.
Thanks.
Regards,
Igor
|