Estimating head-motion and deformations derived from eddy-currents in diffusion MRI data.

Retrospective estimation of head-motion between diffusion-weighted images (DWI) acquired within diffusion MRI (dMRI) experiments renders exceptionally challenging1 for datasets including high-diffusivity (or “high b”) images. These “high b” (b > 1000s/mm2) DWIs enable higher angular resolution, as compared to more traditional diffusion tensor imaging (DTI) schemes. UNDISTORT1 (Using NonDistorted Images to Simulate a Template Of the Registration Target) was the earliest method addressing this issue, by simulating a target DW image without motion or distortion from a DTI (b=1000s/mm2) scan of the same subject. Later, Andersson and Sotiropoulos2 proposed a similar approach (widely available within the FSL eddy tool), by predicting the target DW image to be registered from the remainder of the dMRI dataset and modeled with a Gaussian process. Besides the need for less data, eddy has the advantage of implicitly modeling distortions due to Eddy currents. More recently, Cieslak et al.3 integrated both approaches in SHORELine, by (i) setting up a leave-one-out prediction framework as in eddy; and (ii) replacing eddy’s general-purpose Gaussian process prediction with the SHORE4 diffusion model.

Eddymotion is an open implementation of eddy-current and head-motion correction that builds upon the work of eddy and SHORELine, while generalizing these methods to multiple acquisition schemes (single-shell, multi-shell, and diffusion spectrum imaging) using diffusion models available with DIPY5.