Classification of Findings with Localized Lesions in Fundoscopic Images Using a Regionally Guided CNN

Jaemin Son, Woong Bae, Sangkeun Kim, Sang Jun Park, Kyu Hwan Jung

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

Fundoscopic images are often investigated by ophthalmologists to spot abnormal lesions to make diagnoses. Recent successes of convolutional neural networks are confined to diagnoses of few diseases without proper localization of lesion. In this paper, we propose an efficient annotation method for localizing lesions and a CNN architecture that can classify an individual finding and localize the lesions at the same time. Also, we introduce a new loss function to guide the network to learn meaningful patterns with the guidance of the regional annotations. In experiments, we demonstrate that our network performed better than the widely used network and the guidance loss helps achieve higher AUROC up to 4.1 % and superior localization capability.

Original languageEnglish
Title of host publicationComputational Pathology and Ophthalmic Medical Image Analysis - First International Workshop, COMPAY 2018, and 5th International Workshop, OMIA 2018, Held in Conjunction with MICCAI 2018, Proceedings
EditorsZeike Taylor, Hrvoje Bogunovic, David Snead, Mona K. Garvin, Xin Jan Chen, Francesco Ciompi, Yanwu Xu, Lena Maier-Hein, Mitko Veta, Emanuele Trucco, Danail Stoyanov, Nasir Rajpoot, Jeroen van der Laak, Anne Martel, Stephen McKenna
PublisherSpringer Verlag
Pages176-184
Number of pages9
ISBN (Print)9783030009489
DOIs
StatePublished - 1 Jan 2018
Event1st International Workshop on Computational Pathology, COMPAY 2018 and 5th International Workshop on Ophthalmic Medical Image Analysis, OMIA 2018 Held in Conjunction with MICCAI 2018 - Granada, Spain
Duration: 16 Sep 201820 Sep 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11039 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other1st International Workshop on Computational Pathology, COMPAY 2018 and 5th International Workshop on Ophthalmic Medical Image Analysis, OMIA 2018 Held in Conjunction with MICCAI 2018
CountrySpain
CityGranada
Period16/09/1820/09/18

Fingerprint

Guidance
Annotation
Loss Function
Neural networks
Classify
Neural Networks
Experiments
Demonstrate
Experiment
Architecture

Cite this

Son, J., Bae, W., Kim, S., Park, S. J., & Jung, K. H. (2018). Classification of Findings with Localized Lesions in Fundoscopic Images Using a Regionally Guided CNN. In Z. Taylor, H. Bogunovic, D. Snead, M. K. Garvin, X. J. Chen, F. Ciompi, Y. Xu, L. Maier-Hein, M. Veta, E. Trucco, D. Stoyanov, N. Rajpoot, J. van der Laak, A. Martel, ... S. McKenna (Eds.), Computational Pathology and Ophthalmic Medical Image Analysis - First International Workshop, COMPAY 2018, and 5th International Workshop, OMIA 2018, Held in Conjunction with MICCAI 2018, Proceedings (pp. 176-184). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11039 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-00949-6_21
Son, Jaemin ; Bae, Woong ; Kim, Sangkeun ; Park, Sang Jun ; Jung, Kyu Hwan. / Classification of Findings with Localized Lesions in Fundoscopic Images Using a Regionally Guided CNN. Computational Pathology and Ophthalmic Medical Image Analysis - First International Workshop, COMPAY 2018, and 5th International Workshop, OMIA 2018, Held in Conjunction with MICCAI 2018, Proceedings. editor / Zeike Taylor ; Hrvoje Bogunovic ; David Snead ; Mona K. Garvin ; Xin Jan Chen ; Francesco Ciompi ; Yanwu Xu ; Lena Maier-Hein ; Mitko Veta ; Emanuele Trucco ; Danail Stoyanov ; Nasir Rajpoot ; Jeroen van der Laak ; Anne Martel ; Stephen McKenna. Springer Verlag, 2018. pp. 176-184 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{185a08aaadcd48888caa70bcffc3b505,
title = "Classification of Findings with Localized Lesions in Fundoscopic Images Using a Regionally Guided CNN",
abstract = "Fundoscopic images are often investigated by ophthalmologists to spot abnormal lesions to make diagnoses. Recent successes of convolutional neural networks are confined to diagnoses of few diseases without proper localization of lesion. In this paper, we propose an efficient annotation method for localizing lesions and a CNN architecture that can classify an individual finding and localize the lesions at the same time. Also, we introduce a new loss function to guide the network to learn meaningful patterns with the guidance of the regional annotations. In experiments, we demonstrate that our network performed better than the widely used network and the guidance loss helps achieve higher AUROC up to 4.1 {\%} and superior localization capability.",
author = "Jaemin Son and Woong Bae and Sangkeun Kim and Park, {Sang Jun} and Jung, {Kyu Hwan}",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/978-3-030-00949-6_21",
language = "English",
isbn = "9783030009489",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "176--184",
editor = "Zeike Taylor and Hrvoje Bogunovic and David Snead and Garvin, {Mona K.} and Chen, {Xin Jan} and Francesco Ciompi and Yanwu Xu and Lena Maier-Hein and Mitko Veta and Emanuele Trucco and Danail Stoyanov and Nasir Rajpoot and {van der Laak}, Jeroen and Anne Martel and Stephen McKenna",
booktitle = "Computational Pathology and Ophthalmic Medical Image Analysis - First International Workshop, COMPAY 2018, and 5th International Workshop, OMIA 2018, Held in Conjunction with MICCAI 2018, Proceedings",

}

Son, J, Bae, W, Kim, S, Park, SJ & Jung, KH 2018, Classification of Findings with Localized Lesions in Fundoscopic Images Using a Regionally Guided CNN. in Z Taylor, H Bogunovic, D Snead, MK Garvin, XJ Chen, F Ciompi, Y Xu, L Maier-Hein, M Veta, E Trucco, D Stoyanov, N Rajpoot, J van der Laak, A Martel & S McKenna (eds), Computational Pathology and Ophthalmic Medical Image Analysis - First International Workshop, COMPAY 2018, and 5th International Workshop, OMIA 2018, Held in Conjunction with MICCAI 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11039 LNCS, Springer Verlag, pp. 176-184, 1st International Workshop on Computational Pathology, COMPAY 2018 and 5th International Workshop on Ophthalmic Medical Image Analysis, OMIA 2018 Held in Conjunction with MICCAI 2018, Granada, Spain, 16/09/18. https://doi.org/10.1007/978-3-030-00949-6_21

Classification of Findings with Localized Lesions in Fundoscopic Images Using a Regionally Guided CNN. / Son, Jaemin; Bae, Woong; Kim, Sangkeun; Park, Sang Jun; Jung, Kyu Hwan.

Computational Pathology and Ophthalmic Medical Image Analysis - First International Workshop, COMPAY 2018, and 5th International Workshop, OMIA 2018, Held in Conjunction with MICCAI 2018, Proceedings. ed. / Zeike Taylor; Hrvoje Bogunovic; David Snead; Mona K. Garvin; Xin Jan Chen; Francesco Ciompi; Yanwu Xu; Lena Maier-Hein; Mitko Veta; Emanuele Trucco; Danail Stoyanov; Nasir Rajpoot; Jeroen van der Laak; Anne Martel; Stephen McKenna. Springer Verlag, 2018. p. 176-184 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11039 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

TY - GEN

T1 - Classification of Findings with Localized Lesions in Fundoscopic Images Using a Regionally Guided CNN

AU - Son, Jaemin

AU - Bae, Woong

AU - Kim, Sangkeun

AU - Park, Sang Jun

AU - Jung, Kyu Hwan

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Fundoscopic images are often investigated by ophthalmologists to spot abnormal lesions to make diagnoses. Recent successes of convolutional neural networks are confined to diagnoses of few diseases without proper localization of lesion. In this paper, we propose an efficient annotation method for localizing lesions and a CNN architecture that can classify an individual finding and localize the lesions at the same time. Also, we introduce a new loss function to guide the network to learn meaningful patterns with the guidance of the regional annotations. In experiments, we demonstrate that our network performed better than the widely used network and the guidance loss helps achieve higher AUROC up to 4.1 % and superior localization capability.

AB - Fundoscopic images are often investigated by ophthalmologists to spot abnormal lesions to make diagnoses. Recent successes of convolutional neural networks are confined to diagnoses of few diseases without proper localization of lesion. In this paper, we propose an efficient annotation method for localizing lesions and a CNN architecture that can classify an individual finding and localize the lesions at the same time. Also, we introduce a new loss function to guide the network to learn meaningful patterns with the guidance of the regional annotations. In experiments, we demonstrate that our network performed better than the widely used network and the guidance loss helps achieve higher AUROC up to 4.1 % and superior localization capability.

UR - http://www.scopus.com/inward/record.url?scp=85053901382&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-00949-6_21

DO - 10.1007/978-3-030-00949-6_21

M3 - Conference contribution

SN - 9783030009489

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

SP - 176

EP - 184

BT - Computational Pathology and Ophthalmic Medical Image Analysis - First International Workshop, COMPAY 2018, and 5th International Workshop, OMIA 2018, Held in Conjunction with MICCAI 2018, Proceedings

A2 - Taylor, Zeike

A2 - Bogunovic, Hrvoje

A2 - Snead, David

A2 - Garvin, Mona K.

A2 - Chen, Xin Jan

A2 - Ciompi, Francesco

A2 - Xu, Yanwu

A2 - Maier-Hein, Lena

A2 - Veta, Mitko

A2 - Trucco, Emanuele

A2 - Stoyanov, Danail

A2 - Rajpoot, Nasir

A2 - van der Laak, Jeroen

A2 - Martel, Anne

A2 - McKenna, Stephen

PB - Springer Verlag

ER -

Son J, Bae W, Kim S, Park SJ, Jung KH. Classification of Findings with Localized Lesions in Fundoscopic Images Using a Regionally Guided CNN. In Taylor Z, Bogunovic H, Snead D, Garvin MK, Chen XJ, Ciompi F, Xu Y, Maier-Hein L, Veta M, Trucco E, Stoyanov D, Rajpoot N, van der Laak J, Martel A, McKenna S, editors, Computational Pathology and Ophthalmic Medical Image Analysis - First International Workshop, COMPAY 2018, and 5th International Workshop, OMIA 2018, Held in Conjunction with MICCAI 2018, Proceedings. Springer Verlag. 2018. p. 176-184. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-00949-6_21