Respiratory motion correction for 2D cine cardiac MR images using probabilistic edge maps

Abstract

2D short axis (SA) cine MR images are of crucial importance for the accurate assessment of cardiac anatomy and function. Since SA cine stacks are routinely acquired during multiple breath-holds, different breath-hold positions can cause a misalignment of the heart between different slices, with potential detrimental effects on a variety of clinically relevant measurements (e.g. volume or shape of the left ventricle). In this study, we propose a novel approach to spatially align motion corrupted SA slices in MR image stacks using 3D probabilistic edge maps (PEMs) generated with structured decision forests. In our technique, each 2D SA slice is associated with a 3D PEM outlining the myocardial contours in the same slice as well as in the adjacent one. In-plane spatial misalignment between adjacent slices is then corrected using a registration algorithm applied to the associated PEMs. This approach was tested against a conventional intensity-based registration method on SA cine stacks acquired from 26 healthy subjects, for whom anatomical 3D cardiac images were also available as reference. End-diastolic left ventricular volumes are estimated using a 3D multi-atlas segmentation technique and used to quantify alignment accuracy. The results show that the proposed technique successfully reduces the misalignment between slices and that the registered stacks allow a more accurate volumetric estimation than both the original and the intensity-corrected ones. � 2016 CCAL.