IRUS Total

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

File Description SizeFormat 
08624549.pdfAccepted version9.99 MBAdobe PDFView/Open
08624549.pdf6.07 MBAdobe PDFView/Open
Title: Automatic 3D bi-ventricular segmentation of cardiac images by a shape-refined multi-task deep learning approach
Authors: Duan, J
Bello, G
Schlemper, J
Bai, W
Dawes, TJW
Biffi, C
Marvao, AD
Doumou, G
O'Regan, DP
Rueckert, D
Item Type: Journal Article
Abstract: 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.
Issue Date: Sep-2019
Date of Acceptance: 1-Jan-2019
URI: http://hdl.handle.net/10044/1/63010
DOI: 10.1109/TMI.2019.2894322
ISSN: 0278-0062
Publisher: Institute of Electrical and Electronics Engineers
Start Page: 2151
End Page: 2164
Journal / Book Title: IEEE Transactions on Medical Imaging
Volume: 38
Issue: 9
Copyright Statement: © 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see http://creativecommons.org/licenses/by/4.0/
Sponsor/Funder: Imperial College London
British Heart Foundation
Engineering & Physical Science Research Council (EPSRC)
Imperial College Healthcare NHS Trust- BRC Funding
Imperial College Healthcare NHS Trust- BRC Funding
Engineering & Physical Science Research Council (EPSRC)
Funder's Grant Number: NH/17/1/32725
Keywords: Science & Technology
Life Sciences & Biomedicine
Computer Science, Interdisciplinary Applications
Engineering, Biomedical
Engineering, Electrical & Electronic
Imaging Science & Photographic Technology
Radiology, Nuclear Medicine & Medical Imaging
Computer Science
Deep learning
bi-ventricular CMR segmentation
landmark localization
non-rigid registration
label fusion
multi-atlas segmentation
shape prior
cardiac artifacts
08 Information and Computing Sciences
09 Engineering
Nuclear Medicine & Medical Imaging
Publication Status: Published
Online Publication Date: 2019-01-23
Appears in Collections:Faculty of Engineering
Department of Brain Sciences

Unless otherwise indicated, items in Spiral are protected by copyright and are licensed under a Creative Commons Attribution NonCommercial NoDerivatives License.

Creative Commons