ASDNet: Attention Based Semi-supervised Deep Networks for Medical Image Segmentation

Dong Nie, Yaozong Gao, Li Wang, Dinggang Shen

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

11 Citations (Scopus)

Abstract

Segmentation is a key step for various medical image analysis tasks. Recently, deep neural networks could provide promising solutions for automatic image segmentation. The network training usually involves a large scale of training data with corresponding ground truth label maps. However, it is very challenging to obtain the ground-truth label maps due to the requirement of expertise knowledge and also intensive labor work. To address such challenges, we propose a novel semi-supervised deep learning framework, called “Attention based Semi-supervised Deep Networks” (ASDNet), to fulfill the segmentation tasks in an end-to-end fashion. Specifically, we propose a fully convolutional confidence network to adversarially train the segmentation network. Based on the confidence map from the confidence network, we then propose a region-attention based semi-supervised learning strategy to include the unlabeled data for training. Besides, sample attention mechanism is also explored to improve the network training. Experimental results on real clinical datasets show that our ASDNet can achieve state-of-the-art segmentation accuracy. Further analysis also indicates that our proposed network components contribute most to the improvement of performance.

Original languageEnglish
Title of host publicationMedical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings
EditorsAlejandro F. Frangi, Gabor Fichtinger, Julia A. Schnabel, Carlos Alberola-López, Christos Davatzikos
PublisherSpringer Verlag
Pages370-378
Number of pages9
ISBN (Print)9783030009366
DOIs
Publication statusPublished - 2018 Jan 1
Externally publishedYes
Event21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018 - Granada, Spain
Duration: 2018 Sep 162018 Sep 20

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11073 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018
CountrySpain
CityGranada
Period18/9/1618/9/20

Fingerprint

Medical Image
Image segmentation
Image Segmentation
Labels
Network components
Segmentation
Supervised learning
Confidence
Image analysis
Semi-supervised Learning
Personnel
Medical Image Analysis
Learning Strategies
Expertise
Neural Networks
Training
Requirements
Experimental Results

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Nie, D., Gao, Y., Wang, L., & Shen, D. (2018). ASDNet: Attention Based Semi-supervised Deep Networks for Medical Image Segmentation. In A. F. Frangi, G. Fichtinger, J. A. Schnabel, C. Alberola-López, & C. Davatzikos (Eds.), Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings (pp. 370-378). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11073 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-00937-3_43

ASDNet : Attention Based Semi-supervised Deep Networks for Medical Image Segmentation. / Nie, Dong; Gao, Yaozong; Wang, Li; Shen, Dinggang.

Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. ed. / Alejandro F. Frangi; Gabor Fichtinger; Julia A. Schnabel; Carlos Alberola-López; Christos Davatzikos. Springer Verlag, 2018. p. 370-378 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11073 LNCS).

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

Nie, D, Gao, Y, Wang, L & Shen, D 2018, ASDNet: Attention Based Semi-supervised Deep Networks for Medical Image Segmentation. in AF Frangi, G Fichtinger, JA Schnabel, C Alberola-López & C Davatzikos (eds), Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11073 LNCS, Springer Verlag, pp. 370-378, 21st International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2018, Granada, Spain, 18/9/16. https://doi.org/10.1007/978-3-030-00937-3_43
Nie D, Gao Y, Wang L, Shen D. ASDNet: Attention Based Semi-supervised Deep Networks for Medical Image Segmentation. In Frangi AF, Fichtinger G, Schnabel JA, Alberola-López C, Davatzikos C, editors, Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. Springer Verlag. 2018. p. 370-378. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-00937-3_43
Nie, Dong ; Gao, Yaozong ; Wang, Li ; Shen, Dinggang. / ASDNet : Attention Based Semi-supervised Deep Networks for Medical Image Segmentation. Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings. editor / Alejandro F. Frangi ; Gabor Fichtinger ; Julia A. Schnabel ; Carlos Alberola-López ; Christos Davatzikos. Springer Verlag, 2018. pp. 370-378 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{ff4e132461be4d5bb3f8e415c6847047,
title = "ASDNet: Attention Based Semi-supervised Deep Networks for Medical Image Segmentation",
abstract = "Segmentation is a key step for various medical image analysis tasks. Recently, deep neural networks could provide promising solutions for automatic image segmentation. The network training usually involves a large scale of training data with corresponding ground truth label maps. However, it is very challenging to obtain the ground-truth label maps due to the requirement of expertise knowledge and also intensive labor work. To address such challenges, we propose a novel semi-supervised deep learning framework, called “Attention based Semi-supervised Deep Networks” (ASDNet), to fulfill the segmentation tasks in an end-to-end fashion. Specifically, we propose a fully convolutional confidence network to adversarially train the segmentation network. Based on the confidence map from the confidence network, we then propose a region-attention based semi-supervised learning strategy to include the unlabeled data for training. Besides, sample attention mechanism is also explored to improve the network training. Experimental results on real clinical datasets show that our ASDNet can achieve state-of-the-art segmentation accuracy. Further analysis also indicates that our proposed network components contribute most to the improvement of performance.",
author = "Dong Nie and Yaozong Gao and Li Wang and Dinggang Shen",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/978-3-030-00937-3_43",
language = "English",
isbn = "9783030009366",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "370--378",
editor = "Frangi, {Alejandro F.} and Gabor Fichtinger and Schnabel, {Julia A.} and Carlos Alberola-L{\'o}pez and Christos Davatzikos",
booktitle = "Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings",

}

TY - GEN

T1 - ASDNet

T2 - Attention Based Semi-supervised Deep Networks for Medical Image Segmentation

AU - Nie, Dong

AU - Gao, Yaozong

AU - Wang, Li

AU - Shen, Dinggang

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Segmentation is a key step for various medical image analysis tasks. Recently, deep neural networks could provide promising solutions for automatic image segmentation. The network training usually involves a large scale of training data with corresponding ground truth label maps. However, it is very challenging to obtain the ground-truth label maps due to the requirement of expertise knowledge and also intensive labor work. To address such challenges, we propose a novel semi-supervised deep learning framework, called “Attention based Semi-supervised Deep Networks” (ASDNet), to fulfill the segmentation tasks in an end-to-end fashion. Specifically, we propose a fully convolutional confidence network to adversarially train the segmentation network. Based on the confidence map from the confidence network, we then propose a region-attention based semi-supervised learning strategy to include the unlabeled data for training. Besides, sample attention mechanism is also explored to improve the network training. Experimental results on real clinical datasets show that our ASDNet can achieve state-of-the-art segmentation accuracy. Further analysis also indicates that our proposed network components contribute most to the improvement of performance.

AB - Segmentation is a key step for various medical image analysis tasks. Recently, deep neural networks could provide promising solutions for automatic image segmentation. The network training usually involves a large scale of training data with corresponding ground truth label maps. However, it is very challenging to obtain the ground-truth label maps due to the requirement of expertise knowledge and also intensive labor work. To address such challenges, we propose a novel semi-supervised deep learning framework, called “Attention based Semi-supervised Deep Networks” (ASDNet), to fulfill the segmentation tasks in an end-to-end fashion. Specifically, we propose a fully convolutional confidence network to adversarially train the segmentation network. Based on the confidence map from the confidence network, we then propose a region-attention based semi-supervised learning strategy to include the unlabeled data for training. Besides, sample attention mechanism is also explored to improve the network training. Experimental results on real clinical datasets show that our ASDNet can achieve state-of-the-art segmentation accuracy. Further analysis also indicates that our proposed network components contribute most to the improvement of performance.

UR - http://www.scopus.com/inward/record.url?scp=85053839687&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85053839687&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-00937-3_43

DO - 10.1007/978-3-030-00937-3_43

M3 - Conference contribution

AN - SCOPUS:85053839687

SN - 9783030009366

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

SP - 370

EP - 378

BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2018 - 21st International Conference, 2018, Proceedings

A2 - Frangi, Alejandro F.

A2 - Fichtinger, Gabor

A2 - Schnabel, Julia A.

A2 - Alberola-López, Carlos

A2 - Davatzikos, Christos

PB - Springer Verlag

ER -