Estimation of non-rigid motion is an important task in respiratory and cardiac motion correction. Usually, this problem is formulated in image space via diffusion, parametric-spline or optical flow methods. However, image-based registration can be impaired by aliasing artefacts or by estimating in low image resolution in cases of highly accelerated acquisitions. In this work, we propose a novel deep learning-based non-rigid motion estimation directly in k-space, named LAPNet. The proposed method, inspired by optical flow, is compared against registration in image space and tested for respiratory and cardiac motion as well as different acquisition trajectories providing a generalizable diffeomorphic registration.
This abstract and the presentation materials are available to members only; a login is required.