A Registration-Based Propagation Framework for Automatic Whole Heart Segmentation of Cardiac MRI

Abstract

Abstract-Magnetic resonance (MR) imaging has become a routine modality for the determination of patient cardiac morphology. The extraction of this information can be important for the development of new clinical applications as well as the planning and guidance of cardiac interventional procedures. To avoid inter-and intra-observer variability of manual delineation, it is highly desirable to develop an automatic technique for whole heart segmentation of cardiac magnetic resonance images. However, automating this process is complicated by the limited quality of acquired images and large shape variation of the heart between subjects. In this paper, we propose a fully automatic whole heart segmentation framework based on two new image registration algorithms: the locally affine registration method (LARM) and the free-form deformations with adaptive control point status (ACPS FFDs). LARM provides the correspondence of anatomical substructures such as the four chambers and great vessels of the heart, while the registration using ACPS FFDs refines the local details using a constrained optimization scheme. We validated our proposed segmentation framework on 37 cardiac MR volumes on the end-diastolic phase, displaying a wide diversity of morphology and pathology, and achieved a mean accuracy of 2.14 0.63 mm (rms surface distance) and a maximal error of 4.31 mm.Index Terms-Atlas, cardiac magnetic resonance imaging (MRI), dynamic resampling and distance weighting interpolation (DRAW), free-form deformations, image registration, inverse transformation, locally affine registration, locally affine registration method (LARM), whole heart segmentation.

Multiatlas whole heart segmentation of CT data using conditional entropy for atlas ranking and selection

“…In our previous work, we have demonstrated that the three-level LARM can improve the atlas-to-target registration of cardiac images and consequently improve the segmentation accuracy of the single atlas-based approach. 12 In this work, we further investigate the performance of this registration technique for MAS. Details of the WHS results are reported to provide a benchmark for future research.…”

“…11,38 However, due to the complex shape and large shape variability of the heart, nonrigid registration may generate unrealistic deformation fields due to the poor initial alignment of substructures by a global affine registration. 12 We adopted a hierarchical registration framework, including a global affine registration for heart localization, a LARM for substructure initialization, and a FFD registration for local refinement. Note that in the heart localization, we used the orientation information contained in the original DICOM files of the atlas and the target image for orientation correction before performing intensity-based registration.…”

“…25 This hierarchical framework was specifically designed for the registration of cardiac images. 12 3.A.1.a. Substructure initialization.…”

“…LARM has been proposed in order to tackle the variation coming from the substructures of the heart, by modeling each of them using a locally affine transformation. 12 LARM is computed as follows:…”

Real‐time MRI guidance of cardiac interventions

“…7778 Notably, deformable models are already computational effective (e.g. 4-30sec) with realistic results being granted by the shape constraint; however, the shape models need large databases for training and boundary detection is crucial in these cases.…”