Dear Sjoerd and Julien,
Sjoerd is absolutely right that you should probably run two separate eddies in this case, but for a slightly different reason.
It is not a problem in itself to mix data with different eddy currents in a single eddy run. The eddy currents are estimated separately for each volume as long as one does not use a 2nd level model for the eddy currents. The default behaviour is to not use a 2nd level model.
But it is a problem in the sense that it is likely that the two runs are afflicted by slightly different susceptibility field/distortions. This is because the subject might be positioned differently and/or because the shim might have changed between the runs. Eddy currently only allows for a single susceptibility field, so would not be able to account for that.
It sounds like you have two distinct sets of b0-scans with opposing PE, one for each dataset. So I would recommend recommend running separate topups and eddys for the two data sets.
Jesper
> Hi Julien,
>
> If these were separate acquisitions, the eddy currents are likely to be different so I would run eddy separately as well. If you are 100% confident that your diffusion-weighting timings are the same (i.e., delta and Delta) you could try and see what eddy does. The thing is, in separate acquisitions with different b-values the scanner typically optimises the gradient timings and magnitudes to get TE as short as possible, which means that you will have different gradient timings and thus very likely to have different behaviour.
>
> HTH,
> Sjoerd
> Hi,
>
> I have two diffusion-weighted datasets acquired in the same subject,
> one with 30 direction at b=1000 (+ one b0) and the other with 64
> directions at b=2000 (+ one b0). Phase-encoding direction is AP. I also
> have two datasets with b=0 and PE direction =AP and PA respectively, for
> distortion correction.
>
> My question is whether I can run eddy on both datasets (b=1000 and
> b-=2000) at once (and how) or whether I should run eddy separately for
> each (after running topup on all b0 volumes)
>
> eddy seems to accept only one file at a time as input.
>
> Thanks
>
> Julien
|