Patching Cardiac and Head Motion Artefacts in Diffusion Imaging

Marcel Peter Zwiers¹

Motion artefacts are an important but often ignored problem in diffusion weighted imaging, and can easily corrupt diffusion model estimations. The post-processing method proposed in this paper uses robust tensor estimation techniques and specifically takes the spatio-temporal structure of the most common artefacts, namely from head and cardiac motion, into account. Simulations demonstrate that the method is more robust and accurate than previous methods and, moreover, improves results significantly using real DWI data. The resulting weights are not limited to diffusion tensor imaging but can also be used to estimate higher order models.