Brain MR Image Segmentation in Small Dataset with Adversarial Defense and Task Reorganization

Xuhua Ren, Lichi Zhang, Dongming Wei, Dinggang Shen, Qian Wang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Medical image segmentation is challenging especially in dealing with small dataset of 3D MR images. Encoding the variation of brain anatomical structures from individual subjects cannot be easily achieved, which is further challenged by only a limited number of well labeled subjects for training. In this study, we aim to address the issue of brain MR image segmentation in small dataset. First, concerning the limited number of training images, we adopt adversarial defense to augment the training data and therefore increase the robustness of the network. Second, inspired by the prior knowledge of neural anatomies, we reorganize the segmentation tasks of different regions into several groups in a hierarchical way. Third, the task reorganization extends to the semantic level, as we incorporate an additional object-level classification task to contribute high-order visual features toward the pixel-level segmentation task. In experiments we validate our method by segmenting gray matter, white matter, and several major regions on a challenge dataset. The proposed method with only seven subjects for training can achieve 84.46% of Dice score in the onsite test set.

Original languageEnglish
Title of host publicationMachine Learning in Medical Imaging - 10th International Workshop, MLMI 2019, Held in Conjunction with MICCAI 2019, Proceedings
EditorsHeung-Il Suk, Mingxia Liu, Chunfeng Lian, Pingkun Yan
PublisherSpringer
Pages1-8
Number of pages8
ISBN (Print)9783030326913
DOIs
Publication statusPublished - 2019 Jan 1
Externally publishedYes
Event10th International Workshop on Machine Learning in Medical Imaging, MLMI 2019 held in conjunction with the 22nd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2019 - Shenzhen, China
Duration: 2019 Oct 132019 Oct 13

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11861 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference10th International Workshop on Machine Learning in Medical Imaging, MLMI 2019 held in conjunction with the 22nd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2019
CountryChina
CityShenzhen
Period19/10/1319/10/13

Fingerprint

Image segmentation
Image Segmentation
Brain
Segmentation
Pixels
Semantics
Dice
Anatomy
Medical Image
Test Set
Prior Knowledge
Encoding
Pixel
Experiments
Higher Order
Robustness
Training
Experiment

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Ren, X., Zhang, L., Wei, D., Shen, D., & Wang, Q. (2019). Brain MR Image Segmentation in Small Dataset with Adversarial Defense and Task Reorganization. In H-I. Suk, M. Liu, C. Lian, & P. Yan (Eds.), Machine Learning in Medical Imaging - 10th International Workshop, MLMI 2019, Held in Conjunction with MICCAI 2019, Proceedings (pp. 1-8). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11861 LNCS). Springer. https://doi.org/10.1007/978-3-030-32692-0_1

Brain MR Image Segmentation in Small Dataset with Adversarial Defense and Task Reorganization. / Ren, Xuhua; Zhang, Lichi; Wei, Dongming; Shen, Dinggang; Wang, Qian.

Machine Learning in Medical Imaging - 10th International Workshop, MLMI 2019, Held in Conjunction with MICCAI 2019, Proceedings. ed. / Heung-Il Suk; Mingxia Liu; Chunfeng Lian; Pingkun Yan. Springer, 2019. p. 1-8 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11861 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Ren, X, Zhang, L, Wei, D, Shen, D & Wang, Q 2019, Brain MR Image Segmentation in Small Dataset with Adversarial Defense and Task Reorganization. in H-I Suk, M Liu, C Lian & P Yan (eds), Machine Learning in Medical Imaging - 10th International Workshop, MLMI 2019, Held in Conjunction with MICCAI 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11861 LNCS, Springer, pp. 1-8, 10th International Workshop on Machine Learning in Medical Imaging, MLMI 2019 held in conjunction with the 22nd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2019, Shenzhen, China, 19/10/13. https://doi.org/10.1007/978-3-030-32692-0_1
Ren X, Zhang L, Wei D, Shen D, Wang Q. Brain MR Image Segmentation in Small Dataset with Adversarial Defense and Task Reorganization. In Suk H-I, Liu M, Lian C, Yan P, editors, Machine Learning in Medical Imaging - 10th International Workshop, MLMI 2019, Held in Conjunction with MICCAI 2019, Proceedings. Springer. 2019. p. 1-8. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-32692-0_1
Ren, Xuhua ; Zhang, Lichi ; Wei, Dongming ; Shen, Dinggang ; Wang, Qian. / Brain MR Image Segmentation in Small Dataset with Adversarial Defense and Task Reorganization. Machine Learning in Medical Imaging - 10th International Workshop, MLMI 2019, Held in Conjunction with MICCAI 2019, Proceedings. editor / Heung-Il Suk ; Mingxia Liu ; Chunfeng Lian ; Pingkun Yan. Springer, 2019. pp. 1-8 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{a335f808a39a44c1ac44734242ec1a55,
title = "Brain MR Image Segmentation in Small Dataset with Adversarial Defense and Task Reorganization",
abstract = "Medical image segmentation is challenging especially in dealing with small dataset of 3D MR images. Encoding the variation of brain anatomical structures from individual subjects cannot be easily achieved, which is further challenged by only a limited number of well labeled subjects for training. In this study, we aim to address the issue of brain MR image segmentation in small dataset. First, concerning the limited number of training images, we adopt adversarial defense to augment the training data and therefore increase the robustness of the network. Second, inspired by the prior knowledge of neural anatomies, we reorganize the segmentation tasks of different regions into several groups in a hierarchical way. Third, the task reorganization extends to the semantic level, as we incorporate an additional object-level classification task to contribute high-order visual features toward the pixel-level segmentation task. In experiments we validate our method by segmenting gray matter, white matter, and several major regions on a challenge dataset. The proposed method with only seven subjects for training can achieve 84.46{\%} of Dice score in the onsite test set.",
author = "Xuhua Ren and Lichi Zhang and Dongming Wei and Dinggang Shen and Qian Wang",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/978-3-030-32692-0_1",
language = "English",
isbn = "9783030326913",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer",
pages = "1--8",
editor = "Heung-Il Suk and Mingxia Liu and Chunfeng Lian and Pingkun Yan",
booktitle = "Machine Learning in Medical Imaging - 10th International Workshop, MLMI 2019, Held in Conjunction with MICCAI 2019, Proceedings",

}

TY - GEN

T1 - Brain MR Image Segmentation in Small Dataset with Adversarial Defense and Task Reorganization

AU - Ren, Xuhua

AU - Zhang, Lichi

AU - Wei, Dongming

AU - Shen, Dinggang

AU - Wang, Qian

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Medical image segmentation is challenging especially in dealing with small dataset of 3D MR images. Encoding the variation of brain anatomical structures from individual subjects cannot be easily achieved, which is further challenged by only a limited number of well labeled subjects for training. In this study, we aim to address the issue of brain MR image segmentation in small dataset. First, concerning the limited number of training images, we adopt adversarial defense to augment the training data and therefore increase the robustness of the network. Second, inspired by the prior knowledge of neural anatomies, we reorganize the segmentation tasks of different regions into several groups in a hierarchical way. Third, the task reorganization extends to the semantic level, as we incorporate an additional object-level classification task to contribute high-order visual features toward the pixel-level segmentation task. In experiments we validate our method by segmenting gray matter, white matter, and several major regions on a challenge dataset. The proposed method with only seven subjects for training can achieve 84.46% of Dice score in the onsite test set.

AB - Medical image segmentation is challenging especially in dealing with small dataset of 3D MR images. Encoding the variation of brain anatomical structures from individual subjects cannot be easily achieved, which is further challenged by only a limited number of well labeled subjects for training. In this study, we aim to address the issue of brain MR image segmentation in small dataset. First, concerning the limited number of training images, we adopt adversarial defense to augment the training data and therefore increase the robustness of the network. Second, inspired by the prior knowledge of neural anatomies, we reorganize the segmentation tasks of different regions into several groups in a hierarchical way. Third, the task reorganization extends to the semantic level, as we incorporate an additional object-level classification task to contribute high-order visual features toward the pixel-level segmentation task. In experiments we validate our method by segmenting gray matter, white matter, and several major regions on a challenge dataset. The proposed method with only seven subjects for training can achieve 84.46% of Dice score in the onsite test set.

UR - http://www.scopus.com/inward/record.url?scp=85075643963&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85075643963&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-32692-0_1

DO - 10.1007/978-3-030-32692-0_1

M3 - Conference contribution

AN - SCOPUS:85075643963

SN - 9783030326913

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

SP - 1

EP - 8

BT - Machine Learning in Medical Imaging - 10th International Workshop, MLMI 2019, Held in Conjunction with MICCAI 2019, Proceedings

A2 - Suk, Heung-Il

A2 - Liu, Mingxia

A2 - Lian, Chunfeng

A2 - Yan, Pingkun

PB - Springer

ER -