Dear Michael,
>
> I'm trying to create a NODDI acquisition with a b2500 run with 6 b0s, and a b800 run with 3 b0s. Thus far, the problem I've had is that sometimes, motion between the runs causes problems if I try and do an fslmerge first, and a single run of eddy (i.e. there is uncorrected motion between the two acquisitions). So, instead on data where I've acquired b800 and b2500 runs separately, I am running two different eddy's, then using a combination of flirt to align the different b0s and b-shells.
>
> I am about to start another study, and am thinking of combining the b800 and b2500 into one run to avoid motion between the runs. If I go this route, would it be better to keep the b800 and b2500 shells separate (i.e. acquire 30D b800 and 60D b2500), or intermingle the shells? Given this will be about 14 minutes total, I would expect some drift motion occurring over time, so I am wondering what will best correct for that.
what version of eddy are you using.
The 5.0.9. patch will perform a “post-eddy” alignment across shells by default, and I would hope that would fix any between-shell alignment problems.
My recommendation would be to always mix the shells in one acquisition. If for example your three first volumes were b0 800 2500 ... one would expect little/no movement between those three volumes. In that case I recommend to not do the “post-eddy alignment”, i.e. to turn it off using the --dont_peas option.
If you acquire them like b0 800 800 .... break ... b0 2500 2500 ... then I would recommend leaving the “post-eddy alignment” on.
If there is a big movement, alternatively a big temperature-induced translation in the PE direction, during the break, yes then the --fwhm=10,0,0,0,0 can help.
Jesper
>
> Thank you.
|