Dear Andrew,
> I was trying to implement topup recently by using topup->applytopup. I realize that the most typical and supported use is topup->eddy, but I've already done it the other way.
> I had a couple questions related to motion correction.
> 1. It appears topup motion corrects the input b0s. Does it also attempt to motion correct the rest of the DWIs (ie b=1000s, etc)?
> 2. If so, is there a tool related to topup that rotates bvecs after motion correction? or
> 3. Should I just switch to using topup -> eddy, since eddy seems to output rotated bvecs?
my strong advice would be to use topup->eddy. My intention was always to replace applytopup with something like eddy once we had it. There is nothing in topup that is able to correct for motion/eddy currents/outliers etc in diffusion weighted images.
I should also repeat that the most important aspect of motion correction is precisely -- motion correction. The rotation of the bvecs is a second order correction and has a relatively very small impact on your data. You can easily verify this for yourself. Take a data set with appreciable amounts of movement and run eddy on it. Then analyse
1. Corrected data with rotated bvecs.
2. Corrected data with original bvecs.
3. Original data with rotated bvecs.
You will find that 1 and 2 are very similar, but that 3 will look quite different.
Jesper
> Thank you for taking the time to read!
|