Dear all,
I have a more theoretical question regarding eddy current correction.
I am wondering why the "eddy" command correction needs diffusion directions sampling the whole or half the 3D sphere. I am acquiring diffusion-weighted data in only one direction, alternating forward and reversed directions and with also forward and reversed phase encoding directions (so lots of repetitions and with several different b-values as well). So topup works perfectly but the "eddy" command crashes saying "data are not shelled" or never ends when using the -data_is_shelled flag.
So, to understand why, I read the reference article for the "eddy" command [Andersson et al., 2016, An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging] and - not pretending to understand all details of the article but - it seems that estimations of eddy currents parameters (beta_i) and rigid body (r_i) parameters are performed for each direction independently. Moreover, the algorithm makes the assumption that directions with small angle difference are similar, to reduce the number of directions I suppose...
So could anyone please explain me why the "eddy" command requires diffusion directions sampling the whole or half the 3D sphere and cannot work with only one diffusion directions acquired in the forward and reversed ways (with several b-values)?
Many thanks in advance for your help!
Best wishes,
Simon
|