Hi Stefano,
> What happens when there is movement between the beginning of the acquisition and the end? The field determined by the first top-up pair should no longer be valid....
there is a movement model incorporated into topup. This is a first order model, so it compensates for the movement itself. It does NOT compensate for movement-by-susceptibility interactions (i.e. the changing of the field as a consequence of subject movement). The latter effect is normally small, but exists. There has been some discussion about if we should incorporate it into eddy (not a mistake, it should go in eddy and not topup) , but that will depend on availability of time.
> What if one has multiple b=0 pairs interspersed in the acquisition sequence....is there some value to checking for movement and applying top-up inputs to several subseries of images separately?
This is a possibility, but that would mean multiple eddy runs as well. In the HCP project we have decided against multiple runs and accept the (small) movement-by-susceptibility effects.
Jesper
>
> Stefano Marenco, MD
> NIMH/CBDB
> 10 Center Drive, Bldg 10 room 3C103
> Bethesda MD 20892
> Tel 301 435-8964
> Fax 301 480-7795
> Email: [log in to unmask]
>
>
> -----Original Message-----
> From: Jesper Andersson [mailto:[log in to unmask]]
> Sent: Tuesday, March 05, 2013 5:24 AM
> To: [log in to unmask]
> Subject: Re: [FSL] topup processing time
>
> Hi again Brian,
>
>> If one were to use topup to do all images (b=0 + 256 directions) with their blip reversed "partner", would combining these in one large acquisition file be the best way to do this in topup or would it be best to do each volume (i.e., direction or b=0) separately with it's reversed pair and then combine them all together in the end for input into eddy?
>
> at the moment (and probably for the future) eddy only takes a single topup-input (i.e. a single susceptibility off-resonance field). That means that if you use all pairs to calculate that field, there is till only that single field going in. The advantage of using more than a single pair of images for topup is only SNR, and that advantage rapidly disappears as you add more pairs. No matter how poor your SNR I don't think there is ever a case (in practice) for using more than five pairs as input into topup.
>
>
> As for eddy, note that it is the UNcorrected images that should be used as input to eddy.
>
> Jesper
>
>
>>
>> I understand that this is not the recommended approach, but I am trying to wrap my head around all the various possibilities and their relative impact.
>>
>> Thanks again.
>> Brian
>>
>
|