Boundary-aware fully convolutional network for brain tumor segmentation

Haocheng Shen; Ruixuan Wang; Jianguo Zhang; Stephen J. McKenna

doi:10.1007/978-3-319-66185-8_49

Boundary-aware fully convolutional network for brain tumor segmentation

Haocheng Shen (Lead / Corresponding author), Ruixuan Wang, Jianguo Zhang, Stephen J. McKenna

Computing

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

88 Citations (Scopus)

955 Downloads (Pure)

Abstract

We propose a novel, multi-task, fully convolutional network (FCN) architecture for automatic segmentation of brain tumor. This network extracts multi-level contextual information by concatenating hierarchical feature representations extracted from multimodal MR images along with their symmetric-difference images. It achieves improved segmentation performance by incorporating boundary information directly into the loss function. The proposed method was evaluated on the BRATS13 and BRATS15 datasets and compared with competing methods on the BRATS13 testing set. Segmented tumor boundaries obtained were better than those obtained by single-task FCN and by FCN with CRF. The method is among the most accurate available and has relatively low computational cost at test time.

Original language	English
Title of host publication	Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings
Place of Publication	Switzerland
Publisher	Springer Verlag
Pages	433-441
Number of pages	9
Volume	10434
ISBN (Electronic)	9783319661858
ISBN (Print)	9783319661841
DOIs	https://doi.org/10.1007/978-3-319-66185-8_49
Publication status	Published - 2017
Event	20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017 - Quebec City, Canada Duration: 11 Sept 2017 → 13 Sept 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Publisher	Springer
Volume	10434
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017
Country/Territory	Canada
City	Quebec City
Period	11/09/17 → 13/09/17

Keywords

Deep learning
Multi-task learning
Tumor segmentation

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1007/978-3-319-66185-8_49

Author Accepted Manuscript
The final publication is available at Springer via http://dx.doi.org/10.1007/978-3-319-66185-8_49.
Accepted author manuscript, 2.27 MB

McKenna, Stephen
- Computing - Professor of Computer Vision
Person: Academic

Cite this

Shen, H., Wang, R., Zhang, J., & McKenna, S. J. (2017). Boundary-aware fully convolutional network for brain tumor segmentation. In Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings (Vol. 10434, pp. 433-441). Article 626 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10434). Springer Verlag. https://doi.org/10.1007/978-3-319-66185-8_49

Shen, Haocheng ; Wang, Ruixuan ; Zhang, Jianguo et al. / Boundary-aware fully convolutional network for brain tumor segmentation. Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10434 Switzerland : Springer Verlag, 2017. pp. 433-441 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{6e8509c6ace84c31a298df17cffd3606,

title = "Boundary-aware fully convolutional network for brain tumor segmentation",

abstract = "We propose a novel, multi-task, fully convolutional network (FCN) architecture for automatic segmentation of brain tumor. This network extracts multi-level contextual information by concatenating hierarchical feature representations extracted from multimodal MR images along with their symmetric-difference images. It achieves improved segmentation performance by incorporating boundary information directly into the loss function. The proposed method was evaluated on the BRATS13 and BRATS15 datasets and compared with competing methods on the BRATS13 testing set. Segmented tumor boundaries obtained were better than those obtained by single-task FCN and by FCN with CRF. The method is among the most accurate available and has relatively low computational cost at test time.",

keywords = "Deep learning, Multi-task learning, Tumor segmentation",

author = "Haocheng Shen and Ruixuan Wang and Jianguo Zhang and McKenna, {Stephen J.}",

note = "This work was supported partially by the National Natural Science Foundation of China (No. 61628212).; 20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017 ; Conference date: 11-09-2017 Through 13-09-2017",

year = "2017",

doi = "10.1007/978-3-319-66185-8_49",

language = "English",

isbn = "9783319661841",

volume = "10434",

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

publisher = "Springer Verlag",

pages = "433--441",

booktitle = "Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings",

address = "Germany",

}

Shen, H, Wang, R, Zhang, J & McKenna, SJ 2017, Boundary-aware fully convolutional network for brain tumor segmentation. in Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. vol. 10434, 626, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10434, Springer Verlag, Switzerland, pp. 433-441, 20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017, Quebec City, Canada, 11/09/17. https://doi.org/10.1007/978-3-319-66185-8_49

Boundary-aware fully convolutional network for brain tumor segmentation. / Shen, Haocheng (Lead / Corresponding author); Wang, Ruixuan; Zhang, Jianguo et al.
Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10434 Switzerland: Springer Verlag, 2017. p. 433-441 626 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10434).

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

TY - GEN

T1 - Boundary-aware fully convolutional network for brain tumor segmentation

AU - Shen, Haocheng

AU - Wang, Ruixuan

AU - Zhang, Jianguo

AU - McKenna, Stephen J.

N1 - This work was supported partially by the National Natural Science Foundation of China (No. 61628212).

PY - 2017

Y1 - 2017

N2 - We propose a novel, multi-task, fully convolutional network (FCN) architecture for automatic segmentation of brain tumor. This network extracts multi-level contextual information by concatenating hierarchical feature representations extracted from multimodal MR images along with their symmetric-difference images. It achieves improved segmentation performance by incorporating boundary information directly into the loss function. The proposed method was evaluated on the BRATS13 and BRATS15 datasets and compared with competing methods on the BRATS13 testing set. Segmented tumor boundaries obtained were better than those obtained by single-task FCN and by FCN with CRF. The method is among the most accurate available and has relatively low computational cost at test time.

AB - We propose a novel, multi-task, fully convolutional network (FCN) architecture for automatic segmentation of brain tumor. This network extracts multi-level contextual information by concatenating hierarchical feature representations extracted from multimodal MR images along with their symmetric-difference images. It achieves improved segmentation performance by incorporating boundary information directly into the loss function. The proposed method was evaluated on the BRATS13 and BRATS15 datasets and compared with competing methods on the BRATS13 testing set. Segmented tumor boundaries obtained were better than those obtained by single-task FCN and by FCN with CRF. The method is among the most accurate available and has relatively low computational cost at test time.

KW - Deep learning

KW - Multi-task learning

KW - Tumor segmentation

U2 - 10.1007/978-3-319-66185-8_49

DO - 10.1007/978-3-319-66185-8_49

M3 - Conference contribution

AN - SCOPUS:85029518728

SN - 9783319661841

VL - 10434

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

SP - 433

EP - 441

BT - Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings

PB - Springer Verlag

CY - Switzerland

T2 - 20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017

Y2 - 11 September 2017 through 13 September 2017

ER -

Shen H, Wang R, Zhang J, McKenna SJ. Boundary-aware fully convolutional network for brain tumor segmentation. In Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10434. Switzerland: Springer Verlag. 2017. p. 433-441. 626. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). Epub 2017 Sept 4. doi: 10.1007/978-3-319-66185-8_49

Boundary-aware fully convolutional network for brain tumor segmentation

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

UN SDGs

Access to Document

Fingerprint

Profiles

McKenna, Stephen

Cite this