Automatic 3D bi-ventricular segmentation of cardiac images by a shape-refined multi-task deep learning approach

Deep learning approaches have achieved state-of-the-art performance in cardiac magnetic resonance (CMR) image segmentation. However, most approaches have focused on learning image intensity features for segmentation, whereas the incorporation of anatomical shape priors has received less attention. I...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Duan, Jinming, Bello, Ghalib, Schlemper, Jo, Bai, Wenjia, Dawes, Timothy J W, Biffi, Carlo, de Marvao, Antonio, Doumou, Georgia, O'Regan, Declan P, Rueckert, Daniel
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 13.07.2019
Subjects
Artificial neural networks
Automation
Deep learning
Hypertension
Image segmentation
Landmarks
Magnetic resonance
Mathematical models
Pipelines
Propagation
Robustness (mathematics)
Three dimensional models
Online AccessGet full text

Cover

More Information
Summary:Deep learning approaches have achieved state-of-the-art performance in cardiac magnetic resonance (CMR) image segmentation. However, most approaches have focused on learning image intensity features for segmentation, whereas the incorporation of anatomical shape priors has received less attention. In this paper, we combine a multi-task deep learning approach with atlas propagation to develop a shape-constrained bi-ventricular segmentation pipeline for short-axis CMR volumetric images. The pipeline first employs a fully convolutional network (FCN) that learns segmentation and landmark localisation tasks simultaneously. The architecture of the proposed FCN uses a 2.5D representation, thus combining the computational advantage of 2D FCNs networks and the capability of addressing 3D spatial consistency without compromising segmentation accuracy. Moreover, the refinement step is designed to explicitly enforce a shape constraint and improve segmentation quality. This step is effective for overcoming image artefacts (e.g. due to different breath-hold positions and large slice thickness), which preclude the creation of anatomically meaningful 3D cardiac shapes. The proposed pipeline is fully automated, due to network's ability to infer landmarks, which are then used downstream in the pipeline to initialise atlas propagation. We validate the pipeline on 1831 healthy subjects and 649 subjects with pulmonary hypertension. Extensive numerical experiments on the two datasets demonstrate that our proposed method is robust and capable of producing accurate, high-resolution and anatomically smooth bi-ventricular 3D models, despite the artefacts in input CMR volumes.
ISSN:2331-8422