Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images

Jinming Duan; Jo Schlemper; Wenjia Bai; Timothy J.W. Dawes; Ghalib Bello; Carlo Biffi; Georgia Doumou; Antonio De Marvao; Declan P. O’Regan; Daniel Rueckert

doi:10.1007/978-3-030-04747-4_24

Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images

Jinming Duan^*, Jo Schlemper, Wenjia Bai, Timothy J.W. Dawes, Ghalib Bello, Carlo Biffi, Georgia Doumou, Antonio De Marvao, Declan P. O’Regan, Daniel Rueckert

^*Corresponding author for this work

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

2 Citations (Scopus)

265 Downloads (Pure)

Abstract

In this paper, we combine a network-based method with image registration to develop a shape-based bi-ventricular segmentation tool for short-axis cardiac magnetic resonance (CMR) volumetric images. The method first employs a fully convolutional network (FCN) to learn the segmentation task from manually labelled ground truth CMR volumes. However, due to the presence of image artefacts in the training dataset, the resulting FCN segmentation results are often imperfect. As such, we propose a second step to refine the FCN segmentation. This step involves performing a non-rigid registration with multiple high-resolution bi-ventricular atlases, allowing the explicit shape priors to be inferred. We validate the proposed approach on 1831 healthy subjects and 200 subjects with pulmonary hypertension. Numerical experiments on the two datasets demonstrate that our approach is capable of producing accurate, high-resolution and anatomically smooth bi-ventricular models, despite the artefacts in the input CMR volumes.

Original language	English
Title of host publication	Shape in Medical Imaging
Subtitle of host publication	nternational Workshop, ShapeMI 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 20, 2018, Proceedings
Editors	Hervé Lombaert, Beatriz Paniagua, Bernhard Egger, Marcel Lüthi, Martin Reuter, Christian Wachinger
Publisher	Springer Verlag
Pages	258-267
Number of pages	10
ISBN (Electronic)	978-3-030-04747-4
ISBN (Print)	978-3-030-04746-7
DOIs	https://doi.org/10.1007/978-3-030-04747-4_24
Publication status	Published - 23 Nov 2018
Event	International Workshop on Shape in Medical Imaging, ShapeMI 2018 held in conjunction with 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018 - Granada, Spain Duration: 20 Sept 2018 → 20 Sept 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11167 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	International Workshop on Shape in Medical Imaging, ShapeMI 2018 held in conjunction with 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018
Country/Territory	Spain
City	Granada
Period	20/09/18 → 20/09/18

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/978-3-030-04747-4_24Licence: None: All rights reserved

Jinming_Duan_et_al_Combining_deep_learning_and_shape_priors_for_bi_ventricular_segmentation_Shape_in_Medical_Imaging_2018
Checked for eligibility: 16/05/2019 The final authenticated version is available online at: https://doi.org/10.1007/978-3-030-04747-4_24
Accepted author manuscript, 1.09 MBLicence: None: All rights reserved

Cite this

Duan, J., Schlemper, J., Bai, W., Dawes, T. J. W., Bello, G., Biffi, C., Doumou, G., De Marvao, A., O’Regan, D. P., & Rueckert, D. (2018). Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images. In H. Lombaert, B. Paniagua, B. Egger, M. Lüthi, M. Reuter, & C. Wachinger (Eds.), Shape in Medical Imaging : nternational Workshop, ShapeMI 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 20, 2018, Proceedings (pp. 258-267). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11167 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-04747-4_24

Duan, Jinming ; Schlemper, Jo ; Bai, Wenjia et al. / Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images. Shape in Medical Imaging : nternational Workshop, ShapeMI 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 20, 2018, Proceedings. editor / Hervé Lombaert ; Beatriz Paniagua ; Bernhard Egger ; Marcel Lüthi ; Martin Reuter ; Christian Wachinger. Springer Verlag, 2018. pp. 258-267 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{e36a4a52afb8431482d31840f2e88c57,

title = "Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images",

abstract = "In this paper, we combine a network-based method with image registration to develop a shape-based bi-ventricular segmentation tool for short-axis cardiac magnetic resonance (CMR) volumetric images. The method first employs a fully convolutional network (FCN) to learn the segmentation task from manually labelled ground truth CMR volumes. However, due to the presence of image artefacts in the training dataset, the resulting FCN segmentation results are often imperfect. As such, we propose a second step to refine the FCN segmentation. This step involves performing a non-rigid registration with multiple high-resolution bi-ventricular atlases, allowing the explicit shape priors to be inferred. We validate the proposed approach on 1831 healthy subjects and 200 subjects with pulmonary hypertension. Numerical experiments on the two datasets demonstrate that our approach is capable of producing accurate, high-resolution and anatomically smooth bi-ventricular models, despite the artefacts in the input CMR volumes.",

author = "Jinming Duan and Jo Schlemper and Wenjia Bai and Dawes, {Timothy J.W.} and Ghalib Bello and Carlo Biffi and Georgia Doumou and {De Marvao}, Antonio and O{\textquoteright}Regan, {Declan P.} and Daniel Rueckert",

year = "2018",

month = nov,

day = "23",

doi = "10.1007/978-3-030-04747-4_24",

language = "English",

isbn = "978-3-030-04746-7",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "258--267",

editor = "Herv{\'e} Lombaert and Beatriz Paniagua and Bernhard Egger and Marcel L{\"u}thi and Martin Reuter and Christian Wachinger",

booktitle = "Shape in Medical Imaging",

note = "International Workshop on Shape in Medical Imaging, ShapeMI 2018 held in conjunction with 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018 ; Conference date: 20-09-2018 Through 20-09-2018",

}

Duan, J, Schlemper, J, Bai, W, Dawes, TJW, Bello, G, Biffi, C, Doumou, G, De Marvao, A, O’Regan, DP & Rueckert, D 2018, Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images. in H Lombaert, B Paniagua, B Egger, M Lüthi, M Reuter & C Wachinger (eds), Shape in Medical Imaging : nternational Workshop, ShapeMI 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 20, 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11167 LNCS, Springer Verlag, pp. 258-267, International Workshop on Shape in Medical Imaging, ShapeMI 2018 held in conjunction with 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018, Granada, Spain, 20/09/18. https://doi.org/10.1007/978-3-030-04747-4_24

Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images. / Duan, Jinming; Schlemper, Jo; Bai, Wenjia et al.
Shape in Medical Imaging : nternational Workshop, ShapeMI 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 20, 2018, Proceedings. ed. / Hervé Lombaert; Beatriz Paniagua; Bernhard Egger; Marcel Lüthi; Martin Reuter; Christian Wachinger. Springer Verlag, 2018. p. 258-267 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11167 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images

AU - Duan, Jinming

AU - Schlemper, Jo

AU - Bai, Wenjia

AU - Dawes, Timothy J.W.

AU - Bello, Ghalib

AU - Biffi, Carlo

AU - Doumou, Georgia

AU - De Marvao, Antonio

AU - O’Regan, Declan P.

AU - Rueckert, Daniel

PY - 2018/11/23

Y1 - 2018/11/23

N2 - In this paper, we combine a network-based method with image registration to develop a shape-based bi-ventricular segmentation tool for short-axis cardiac magnetic resonance (CMR) volumetric images. The method first employs a fully convolutional network (FCN) to learn the segmentation task from manually labelled ground truth CMR volumes. However, due to the presence of image artefacts in the training dataset, the resulting FCN segmentation results are often imperfect. As such, we propose a second step to refine the FCN segmentation. This step involves performing a non-rigid registration with multiple high-resolution bi-ventricular atlases, allowing the explicit shape priors to be inferred. We validate the proposed approach on 1831 healthy subjects and 200 subjects with pulmonary hypertension. Numerical experiments on the two datasets demonstrate that our approach is capable of producing accurate, high-resolution and anatomically smooth bi-ventricular models, despite the artefacts in the input CMR volumes.

AB - In this paper, we combine a network-based method with image registration to develop a shape-based bi-ventricular segmentation tool for short-axis cardiac magnetic resonance (CMR) volumetric images. The method first employs a fully convolutional network (FCN) to learn the segmentation task from manually labelled ground truth CMR volumes. However, due to the presence of image artefacts in the training dataset, the resulting FCN segmentation results are often imperfect. As such, we propose a second step to refine the FCN segmentation. This step involves performing a non-rigid registration with multiple high-resolution bi-ventricular atlases, allowing the explicit shape priors to be inferred. We validate the proposed approach on 1831 healthy subjects and 200 subjects with pulmonary hypertension. Numerical experiments on the two datasets demonstrate that our approach is capable of producing accurate, high-resolution and anatomically smooth bi-ventricular models, despite the artefacts in the input CMR volumes.

UR - http://www.scopus.com/inward/record.url?scp=85057387021&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-04747-4_24

DO - 10.1007/978-3-030-04747-4_24

M3 - Conference contribution

AN - SCOPUS:85057387021

SN - 978-3-030-04746-7

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 258

EP - 267

BT - Shape in Medical Imaging

A2 - Lombaert, Hervé

A2 - Paniagua, Beatriz

A2 - Egger, Bernhard

A2 - Lüthi, Marcel

A2 - Reuter, Martin

A2 - Wachinger, Christian

PB - Springer Verlag

T2 - International Workshop on Shape in Medical Imaging, ShapeMI 2018 held in conjunction with 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018

Y2 - 20 September 2018 through 20 September 2018

ER -

Duan J, Schlemper J, Bai W, Dawes TJW, Bello G, Biffi C et al. Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images. In Lombaert H, Paniagua B, Egger B, Lüthi M, Reuter M, Wachinger C, editors, Shape in Medical Imaging : nternational Workshop, ShapeMI 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 20, 2018, Proceedings. Springer Verlag. 2018. p. 258-267. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-04747-4_24

Combining deep learning and shape priors for bi-ventricular segmentation of volumetric cardiac magnetic resonance images

Abstract

Publication series

Conference

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this