Dear Michael,
> Just curious if there was a followup on which columns are the rotation angles. The new eddy tool does fabulous at correcting higher order eddy currents at multiple b-values. It would be great to correct the b-vectors for motion. Assuming the angles are in radians, they must be the second set of three columns, as the first set goes as high as 2.0 in a dataset I have (which could represent movement of 2mm but obviously not rotation of 2 radians).
>
> Presumably one could translate columns 4-6 into a b-vector rotation matrix, spending a lot of time of course making sure - signs are in the right place. It would be ideal to ultimately have a script to translate the eddy_parameters log from eddy into an ecclog format.
I expect that in the fulness of time we will do so, but at the moment it is less of a priority than some other improvements we are making to eddy.
My personal take on the whole "rotation of bvecs" business is that it is a very small error/correction. When someone has moved enough for this to become an issue I suspect the data will be pretty rubbish anyway. But due to popular demand we will do it once we reach it on the list.
Meanwhile, columns 4-6 are the rotations in radians and in the expected order (x-y-z).
Jesper
|