Hi Torsten and Stam,
I believe that if you do correct for motion and you have also rotation artefacts then you also need to correct the bvecs. Does this work with the eddy_current tool?
Cheers,
Eleftherios
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Stamatios Sotiropoulos
Sent: 27 July 2010 15:23
To: [log in to unmask]
Subject: Re: [FSL] eddy current correction in DTI
Hi Torsten,
the eddy current distortions consist of translations, scalings and shears. Motion is rotations and translations, so with an affine transformation you capture all of them. This is what eddy correct does, applies an affine transformation to register all images to a reference.
If you do an ex-vivo spin-echo (non-EPI readout) scan, geometric distortions are minimal and assuming that the tissue is well stabilized in the scanner, you also do not have motion. So, you should not need to do any correction. This is also proposed in (Dyrby et al, Human Brain Mapping, 2010). If you however have a reason to run eddy_correct, you should check the result by viewing the timeseries in movie mode in FSLView, as Matt suggested.
Cheers,
Stam
On 27 Jul 2010, at 15:10, Torsten Ruest wrote:
> Sorry, my bad... For some reason I thought the eddy current correction step consists of separate 2 processes, the registration for movement, and something else to correct for distortions (here I thought of something different to registration...). At least I know that now!
>
> Thanks for your help,
>
> Torsten
>
|