Very useful, thanks.
Niklas Lenfeldt, PhD, MSc, MAJ SWE A
Clinical Neuroscience, Umeå university
Phone: +46 70 609 66 87
Email: [log in to unmask]
-----Ursprungligt meddelande-----
Från: FSL - FMRIB's Software Library [mailto:[log in to unmask]] För Jesper Andersson
Skickat: den 26 oktober 2015 11:04
Till: [log in to unmask]
Ämne: Re: [FSL] eddy v. eddy_correct
Dear Niklas,
just to add to what Barbara already said,
1. Yes, Barbara is absolutely right that one doesn't need to have run topup to run eddy. Eddy will still do an equally good job of correcting for subject movement and EC distortions. Our experience is too that it does a much better job than eddy_correct.
2. Ideally you should have sampled on the whole sphere and have >20 directions. If you haven't sampled on the whole sphere it is still worth testing to run it on your data, and then I would recommend using --slm=linear. Running a movie loop in fslview is a very good way of detecting if it has worked or not.
3. There are two more papers coming out that both contain some comparisons to eddy_correct. One is on "pre-proof" early view http://www.sciencedirect.com/science/article/pii/S1053811915009209 and the other one is by Mark Graham et al., and will hopefully appear in early view (NeuroImage) soon.
Jesper
On 25 Oct 2015, at 19:08, Barbara Kreilkamp <[log in to unmask]> wrote:
> Hi Niklas,
>
> Have had some limited experience with this comparison, but there is actually a publication on this, comparing eddy with topup and eddy_correct
>
> http://www.plosone.org/article/fetchObject.action?uri=info:doi/10.1371/journal.pone.0112411&representation=PDF
>
> In my dataset I have seen that eddy_correct can introduce motion in participants that have not moved a lot. And just visually I am happier with eddy's (even without topup) performance than with eddy_correct as it seems to align the DWI much better.
>
> If you can use eddy (if your bvectors were sampled on the whole sphere [check this with MATLAB commands mentioned here: http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/EDDY ], it would be good to do so.
>
> Also, perhaps you have access to a T2w-image?
> You might be able to follow this post on how to use the T2-w image in topup: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A3=ind1303&L=FSL&E=quoted-printable&P=4254046&B=--B_3446959459_95180&T=text%2Fhtml;%20charset=UTF-8&pending=
>
> Cheers,
> Barbara
>
>
>
> On 25/10/2015 18:32, Niklas Lenfeldt wrote:
>> I have data that I previously applied eddy_correct to. Unfortunately, there are no repetitive b-0 volumes with reversed phase encoding, which, as I understand it, makes topup inapplicable.
>>
>> Is there any advantage to still apply eddy (although there is only one phase encoding) compared to eddy_correct (which already has been applied)?
>>
>> Best
>>
>> /Nick
|