Meeting Banner
Abstract #2426

A Comprehensive Gaussian Process Framework for Correcting Distortions and Movements in Diffusion Images

Jesper L. R. Andersson1, Junquian Xu2, Essa Yacoub2, Edward Auerbach2, Steen Moeller2, Kamil Ugurbil2

1FMRIB, Oxford, Oxfordshire, United Kingdom; 2Center for Magnetic Resonance Research (CMRR), University of Minnesota, Minnesota, Minneapolis, United States

Registration based approaches to correcting for eddy current distortions and movements are complicated by the images containing different information. Possible solutions to this is i) To collect pairs of dwis with opposite polarity diffusion gradients, ii) To collect pairs of dwis with opposite polarity phase-encoding or iii) To register observed dwis to model based predictions. We present a method that include any and all of those sources of information depending on how the data was collected. It utilises Gaussian Processes to predict dwis and to predict eddy currents for one acquisition given other acquisitions.

Abstract PDF Poster PDF

Keywords

able abstract acquisition acquisitions addition allow allows ambiguity angular assuming assumption back band bias blue bottom butt central channel coil collected comb combined contrast correcting denote designed despite dictions difficult diffusion distortions drift eddy even field fields floor framework functions funded gradient gradients grants half heart hence highly human ideal induced inform isotropic issues liable like linear loaded look made make maker mates model mono move movement movements next noise note operating opposite oxford parameterized part partly perspective piloting polar polarity precision predict prediction predictions problematic process profile project reduce reduction registration related reload removing respectively sampled scheme severe shells slice space starts suffer suggested supplied susceptibility tensor thanks training transform translates updated utilizes variable volume