Imperial College London
Portrait Picture

ProfessorDeclanO'Regan

Faculty of Medicine, Institute of Clinical Sciences

Professor of Imaging Sciences
 
 
 
//

Contact

 

+44 (0)20 3313 1510declan.oregan

 
 
//

Location

 

Imaging Sciences DepartmentHammersmith HospitalHammersmith Campus

//

Summary

 

Publications

Citation

BibTex format

@article{Duan:2019:10.1109/TMI.2019.2894322,
author = {Duan, J and Bello, G and Schlemper, J and Bai, W and Dawes, TJW and Biffi, C and Marvao, AD and Doumou, G and O'Regan, DP and Rueckert, D},
doi = {10.1109/TMI.2019.2894322},
journal = {IEEE Transactions on Medical Imaging},
pages = {2151--2164},
title = {Automatic 3D bi-ventricular segmentation of cardiac images by a shape-refined multi-task deep learning approach},
url = {http://dx.doi.org/10.1109/TMI.2019.2894322},
volume = {38},
year = {2019}
}
Download

RIS format (EndNote, RefMan)

TY  - JOUR
AB  - Deep learning approaches have achieved state-of-the-art performance incardiac 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-ventricular3D models, despite the artefacts in input CMR volumes.
AU  - Duan,J
AU  - Bello,G
AU  - Schlemper,J
AU  - Bai,W
AU  - Dawes,TJW
AU  - Biffi,C
AU  - Marvao,AD
AU  - Doumou,G
AU  - O'Regan,DP
AU  - Rueckert,D
DO  - 10.1109/TMI.2019.2894322
EP  - 2164
PY  - 2019///
SN  - 0278-0062
SP  - 2151
TI  - Automatic 3D bi-ventricular segmentation of cardiac images by a shape-refined multi-task deep learning approach
T2  - IEEE Transactions on Medical Imaging
UR  - http://dx.doi.org/10.1109/TMI.2019.2894322
UR  - http://arxiv.org/abs/1808.08578v2
UR  - https://ieeexplore.ieee.org/document/8624549
UR  - http://hdl.handle.net/10044/1/63010
VL  - 38
ER  -
Download