Ensemble of deep convolutional neural networks for prognosis of ischemic stroke

Youngwon Choi; Yongchan Kwon; Hanbyul Lee; Beom Joon Kim; Myunghee Cho Paik; Joong Ho Won

doi:10.1007/978-3-319-55524-9_22

Ensemble of deep convolutional neural networks for prognosis of ischemic stroke

Youngwon Choi, Yongchan Kwon, Hanbyul Lee, Beom Joon Kim, Myunghee Cho Paik, Joong Ho Won

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

38 Scopus citations

Abstract

We propose an ensemble of deep neural networks for the two tasks of automated prognosis of post-treatment ischemic stroke, as imposed by the ISLES 2016 Challenge. For lesion outcome prediction, we employ an ensemble of three-dimensional multiscale residual U-Net and a fully convolutional network, trained using image patches. In order to handle class imbalance, we devise a multi-step training strategy. For clinical outcome prediction, we combine a convolutional neural network (CNN) and a logistic regression model. To overcome the small sample size and the need for whole brain image, we use the CNN trained using patches as a feature extractor and trained a shallow network based on the extracted features. Our ensemble approach demonstrated an appealing performance on both problems, and is ranked among the top entries in the Challenge.

Original language	English
Title of host publication	Brainlesion
Subtitle of host publication	Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - Second International Workshop, BrainLes 2016, with the Challenges on BRATS, ISLES and mTOP 2016 Held in Conjunction with MICCAI 2016, Revised Selected Papers
Editors	Bjoern Menze, Mauricio Reyes, Alessandro Crimi, Oskar Maier, Stefan Winzeck, Heinz Handels
Publisher	Springer Verlag
Pages	231-243
Number of pages	13
ISBN (Print)	9783319555232
DOIs	https://doi.org/10.1007/978-3-319-55524-9_22
State	Published - 2016
Event	2nd International Workshop on Brain Lesion, BrainLes 2016, with the challenges on Brain Tumor Segmentation BRATS, Ischemic Stroke Lesion Image Segmentation ISLES, and the Mild Traumatic Brain Injury Outcome Prediction mTOP held in conjunction with the International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016 - Athens, Greece Duration: 17 Oct 2016 → 17 Oct 2016

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10154 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	2nd International Workshop on Brain Lesion, BrainLes 2016, with the challenges on Brain Tumor Segmentation BRATS, Ischemic Stroke Lesion Image Segmentation ISLES, and the Mild Traumatic Brain Injury Outcome Prediction mTOP held in conjunction with the International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016
Country/Territory	Greece
City	Athens
Period	17/10/16 → 17/10/16

Bibliographical note

Publisher Copyright:
© Springer International Publishing AG 2016.

Keywords

3D convolutional kernels
Class imbalance
Ensemble
Fully convolutional networks
Multi-phase training
Patchwise learning
U-Net

Access to Document

10.1007/978-3-319-55524-9_22

Cite this

Choi, Y., Kwon, Y., Lee, H., Kim, B. J., Paik, M. C., & Won, J. H. (2016). Ensemble of deep convolutional neural networks for prognosis of ischemic stroke. In B. Menze, M. Reyes, A. Crimi, O. Maier, S. Winzeck, & H. Handels (Eds.), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - Second International Workshop, BrainLes 2016, with the Challenges on BRATS, ISLES and mTOP 2016 Held in Conjunction with MICCAI 2016, Revised Selected Papers (pp. 231-243). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10154 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-55524-9_22

Choi, Youngwon ; Kwon, Yongchan ; Lee, Hanbyul et al. / Ensemble of deep convolutional neural networks for prognosis of ischemic stroke. Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - Second International Workshop, BrainLes 2016, with the Challenges on BRATS, ISLES and mTOP 2016 Held in Conjunction with MICCAI 2016, Revised Selected Papers. editor / Bjoern Menze ; Mauricio Reyes ; Alessandro Crimi ; Oskar Maier ; Stefan Winzeck ; Heinz Handels. Springer Verlag, 2016. pp. 231-243 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{a9f1d8f0d31d4a0ca9e3c23c9654a4c8,

title = "Ensemble of deep convolutional neural networks for prognosis of ischemic stroke",

abstract = "We propose an ensemble of deep neural networks for the two tasks of automated prognosis of post-treatment ischemic stroke, as imposed by the ISLES 2016 Challenge. For lesion outcome prediction, we employ an ensemble of three-dimensional multiscale residual U-Net and a fully convolutional network, trained using image patches. In order to handle class imbalance, we devise a multi-step training strategy. For clinical outcome prediction, we combine a convolutional neural network (CNN) and a logistic regression model. To overcome the small sample size and the need for whole brain image, we use the CNN trained using patches as a feature extractor and trained a shallow network based on the extracted features. Our ensemble approach demonstrated an appealing performance on both problems, and is ranked among the top entries in the Challenge.",

keywords = "3D convolutional kernels, Class imbalance, Ensemble, Fully convolutional networks, Multi-phase training, Patchwise learning, U-Net",

author = "Youngwon Choi and Yongchan Kwon and Hanbyul Lee and Kim, {Beom Joon} and Paik, {Myunghee Cho} and Won, {Joong Ho}",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG 2016.; 2nd International Workshop on Brain Lesion, BrainLes 2016, with the challenges on Brain Tumor Segmentation BRATS, Ischemic Stroke Lesion Image Segmentation ISLES, and the Mild Traumatic Brain Injury Outcome Prediction mTOP held in conjunction with the International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016 ; Conference date: 17-10-2016 Through 17-10-2016",

year = "2016",

doi = "10.1007/978-3-319-55524-9_22",

language = "English",

isbn = "9783319555232",

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

publisher = "Springer Verlag",

pages = "231--243",

editor = "Bjoern Menze and Mauricio Reyes and Alessandro Crimi and Oskar Maier and Stefan Winzeck and Heinz Handels",

booktitle = "Brainlesion",

}

Choi, Y, Kwon, Y, Lee, H, Kim, BJ, Paik, MC & Won, JH 2016, Ensemble of deep convolutional neural networks for prognosis of ischemic stroke. in B Menze, M Reyes, A Crimi, O Maier, S Winzeck & H Handels (eds), Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - Second International Workshop, BrainLes 2016, with the Challenges on BRATS, ISLES and mTOP 2016 Held in Conjunction with MICCAI 2016, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10154 LNCS, Springer Verlag, pp. 231-243, 2nd International Workshop on Brain Lesion, BrainLes 2016, with the challenges on Brain Tumor Segmentation BRATS, Ischemic Stroke Lesion Image Segmentation ISLES, and the Mild Traumatic Brain Injury Outcome Prediction mTOP held in conjunction with the International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016, Athens, Greece, 17/10/16. https://doi.org/10.1007/978-3-319-55524-9_22

Ensemble of deep convolutional neural networks for prognosis of ischemic stroke. / Choi, Youngwon; Kwon, Yongchan; Lee, Hanbyul et al.
Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - Second International Workshop, BrainLes 2016, with the Challenges on BRATS, ISLES and mTOP 2016 Held in Conjunction with MICCAI 2016, Revised Selected Papers. ed. / Bjoern Menze; Mauricio Reyes; Alessandro Crimi; Oskar Maier; Stefan Winzeck; Heinz Handels. Springer Verlag, 2016. p. 231-243 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10154 LNCS).

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

TY - GEN

T1 - Ensemble of deep convolutional neural networks for prognosis of ischemic stroke

AU - Choi, Youngwon

AU - Kwon, Yongchan

AU - Lee, Hanbyul

AU - Kim, Beom Joon

AU - Paik, Myunghee Cho

AU - Won, Joong Ho

N1 - Publisher Copyright: © Springer International Publishing AG 2016.

PY - 2016

Y1 - 2016

N2 - We propose an ensemble of deep neural networks for the two tasks of automated prognosis of post-treatment ischemic stroke, as imposed by the ISLES 2016 Challenge. For lesion outcome prediction, we employ an ensemble of three-dimensional multiscale residual U-Net and a fully convolutional network, trained using image patches. In order to handle class imbalance, we devise a multi-step training strategy. For clinical outcome prediction, we combine a convolutional neural network (CNN) and a logistic regression model. To overcome the small sample size and the need for whole brain image, we use the CNN trained using patches as a feature extractor and trained a shallow network based on the extracted features. Our ensemble approach demonstrated an appealing performance on both problems, and is ranked among the top entries in the Challenge.

AB - We propose an ensemble of deep neural networks for the two tasks of automated prognosis of post-treatment ischemic stroke, as imposed by the ISLES 2016 Challenge. For lesion outcome prediction, we employ an ensemble of three-dimensional multiscale residual U-Net and a fully convolutional network, trained using image patches. In order to handle class imbalance, we devise a multi-step training strategy. For clinical outcome prediction, we combine a convolutional neural network (CNN) and a logistic regression model. To overcome the small sample size and the need for whole brain image, we use the CNN trained using patches as a feature extractor and trained a shallow network based on the extracted features. Our ensemble approach demonstrated an appealing performance on both problems, and is ranked among the top entries in the Challenge.

KW - 3D convolutional kernels

KW - Class imbalance

KW - Ensemble

KW - Fully convolutional networks

KW - Multi-phase training

KW - Patchwise learning

KW - U-Net

UR - http://www.scopus.com/inward/record.url?scp=85018709553&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-55524-9_22

DO - 10.1007/978-3-319-55524-9_22

M3 - Conference contribution

AN - SCOPUS:85018709553

SN - 9783319555232

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

SP - 231

EP - 243

BT - Brainlesion

A2 - Menze, Bjoern

A2 - Reyes, Mauricio

A2 - Crimi, Alessandro

A2 - Maier, Oskar

A2 - Winzeck, Stefan

A2 - Handels, Heinz

PB - Springer Verlag

T2 - 2nd International Workshop on Brain Lesion, BrainLes 2016, with the challenges on Brain Tumor Segmentation BRATS, Ischemic Stroke Lesion Image Segmentation ISLES, and the Mild Traumatic Brain Injury Outcome Prediction mTOP held in conjunction with the International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016

Y2 - 17 October 2016 through 17 October 2016

ER -

Choi Y, Kwon Y, Lee H, Kim BJ, Paik MC, Won JH. Ensemble of deep convolutional neural networks for prognosis of ischemic stroke. In Menze B, Reyes M, Crimi A, Maier O, Winzeck S, Handels H, editors, Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries - Second International Workshop, BrainLes 2016, with the Challenges on BRATS, ISLES and mTOP 2016 Held in Conjunction with MICCAI 2016, Revised Selected Papers. Springer Verlag. 2016. p. 231-243. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-55524-9_22

Ensemble of deep convolutional neural networks for prognosis of ischemic stroke

Abstract

Publication series

Other

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this