Dear Niklas,
>
> Dear ladies and gentlemen,
> I want to ask wether topup uses some kind of wrap-around (periodic) extrapolation similar to eddy?
yes and no. The images themselves are periodically interpolated in the PE-direction. So if an image is wrapped topup can fix that by translating the image until it has all been “wrapped back”. But the estimated field is not periodic (it wouldn’t really make sense).
> We have a pair of images A->P and P->A, in the P->A image there is a bit of phase wrap-around due to the distortion.
> The result is still good, we use it to correct GE A->P images with applytopup's jacobian-method.
> Should we correct the wrap-around before topup?
If I have images with severe wrapping I always play it safe and wrap them back manually before I pass them to topup. The reason for that is precisely because the field is not periodic, which in practice means that the “effective” field can have a discontinuity at the image border. But I admit that I am a fusspot.
Jesper
> Thank you in advance,
> Niklas Förster
|