Deep learning and ensemble method for optic disc and cup segmentation

Jongwoo Kim; Loc Tran; Tunde Peto; Emily Y. Chew

doi:10.1109/CIBCB55180.2022.9863022

Deep learning and ensemble method for optic disc and cup segmentation

Jongwoo Kim, Loc Tran, Tunde Peto, Emily Y. Chew

School of Medicine, Dentistry and Biomedical Sciences

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

6 Citations (Scopus)

Abstract

Glaucoma is a chronic retinal disease that gradually damages the optic nerve. It is a leading cause of irreversible loss of vision. In ophthalmic fundus images, the cup to optic disc ratio measured around the optic nerve is a key measure used to screen for glaucomatous damages. Unfortunately, there is high subjectivity among ophthalmologists in estimating this ratio due to challenges in making reliable disc and cup measurements. To minimize this, we propose an automatic method using deep learning and ensemble method to segment the optic disc and cup. The proposed method comprises two steps. The region of interest (ROI), where optic disc is centered, is detected from a fundus image, following which the optic disc and cup are segmented from the ROI. Mask R-CNN algorithm is used to estimate the ROI, and two ensemble models based on three fully convolutional networks are used for the segmentation of optic disc and cup in parallel. The proposed method is trained and evaluated using the RIGA dataset that contains 750 fundus images and the REFUGE database containing 400 fundus images. The results demonstrate that the proposed method has a better performance compared with the current state-of-the-art algorithms. Our best segmentation results for optic disc shows 0.9303 Jaccard Index (JI) and 0.9635 Dice Coefficient (DC). The best segmentation results for cup shows 0.8096 JI and 0.8915 DC. The average cup to optic disc ratio error shows 0.0429.

Original language	English
Title of host publication	2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB): proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781665484626
ISBN (Print)	9781665484633
DOIs	https://doi.org/10.1109/CIBCB55180.2022.9863022
Publication status	Published - 26 Aug 2022
Event	2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology, CIBCB 2022 - Ottawa, Canada Duration: 15 Aug 2022 → 17 Aug 2022

Publication series

Name	IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology: proceedings

Conference

Conference	2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology, CIBCB 2022
Country/Territory	Canada
City	Ottawa
Period	15/08/2022 → 17/08/2022

Bibliographical note

Funding Information:
ACKNOWLEDGMENT This research/work was supported in part by the Lister Hill National Center for Biomedical Communications of the National Library of Medicine (NLM), National Institutes of Health. We thank the MESSIDOR program, Magrabi Eye Center in Saudi Arabia and Bin Rushed Ophthalmic center in Saudi Arabia for providing the RIGA dataset. We thank the team of Jose Ignacio Orlando for providing REFUGE dataset. In addition, we acknowledge the help of the National Eye Institute staff in providing additional fundus images (AREDS dataset) to extend the current work.

Publisher Copyright:
© 2022 IEEE.

Keywords

Cup
Deep Learning
Ensemble Method
Fully Convolutional Network (FCN)
Fundus Image
Glaucoma
Optic Disc

ASJC Scopus subject areas

Artificial Intelligence
Information Systems
Information Systems and Management
Computational Mathematics
Health Informatics

Access to Document

10.1109/CIBCB55180.2022.9863022Licence: Unspecified

Cite this

Kim, J., Tran, L., Peto, T., & Chew, E. Y. (2022). Deep learning and ensemble method for optic disc and cup segmentation. In 2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB): proceedings (IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology: proceedings). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CIBCB55180.2022.9863022

Kim, Jongwoo ; Tran, Loc ; Peto, Tunde et al. / Deep learning and ensemble method for optic disc and cup segmentation. 2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB): proceedings. Institute of Electrical and Electronics Engineers Inc., 2022. (IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology: proceedings).

@inproceedings{7f4c765cd0794ca394806e62eca26d91,

title = "Deep learning and ensemble method for optic disc and cup segmentation",

abstract = "Glaucoma is a chronic retinal disease that gradually damages the optic nerve. It is a leading cause of irreversible loss of vision. In ophthalmic fundus images, the cup to optic disc ratio measured around the optic nerve is a key measure used to screen for glaucomatous damages. Unfortunately, there is high subjectivity among ophthalmologists in estimating this ratio due to challenges in making reliable disc and cup measurements. To minimize this, we propose an automatic method using deep learning and ensemble method to segment the optic disc and cup. The proposed method comprises two steps. The region of interest (ROI), where optic disc is centered, is detected from a fundus image, following which the optic disc and cup are segmented from the ROI. Mask R-CNN algorithm is used to estimate the ROI, and two ensemble models based on three fully convolutional networks are used for the segmentation of optic disc and cup in parallel. The proposed method is trained and evaluated using the RIGA dataset that contains 750 fundus images and the REFUGE database containing 400 fundus images. The results demonstrate that the proposed method has a better performance compared with the current state-of-the-art algorithms. Our best segmentation results for optic disc shows 0.9303 Jaccard Index (JI) and 0.9635 Dice Coefficient (DC). The best segmentation results for cup shows 0.8096 JI and 0.8915 DC. The average cup to optic disc ratio error shows 0.0429.",

keywords = "Cup, Deep Learning, Ensemble Method, Fully Convolutional Network (FCN), Fundus Image, Glaucoma, Optic Disc",

author = "Jongwoo Kim and Loc Tran and Tunde Peto and Chew, {Emily Y.}",

note = "Funding Information: ACKNOWLEDGMENT This research/work was supported in part by the Lister Hill National Center for Biomedical Communications of the National Library of Medicine (NLM), National Institutes of Health. We thank the MESSIDOR program, Magrabi Eye Center in Saudi Arabia and Bin Rushed Ophthalmic center in Saudi Arabia for providing the RIGA dataset. We thank the team of Jose Ignacio Orlando for providing REFUGE dataset. In addition, we acknowledge the help of the National Eye Institute staff in providing additional fundus images (AREDS dataset) to extend the current work. Publisher Copyright: {\textcopyright} 2022 IEEE.; 2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology, CIBCB 2022 ; Conference date: 15-08-2022 Through 17-08-2022",

year = "2022",

month = aug,

day = "26",

doi = "10.1109/CIBCB55180.2022.9863022",

language = "English",

isbn = "9781665484633",

series = "IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology: proceedings",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB): proceedings",

address = "United States",

}

Kim, J, Tran, L, Peto, T & Chew, EY 2022, Deep learning and ensemble method for optic disc and cup segmentation. in 2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB): proceedings. IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology: proceedings, Institute of Electrical and Electronics Engineers Inc., 2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology, CIBCB 2022, Ottawa, Canada, 15/08/2022. https://doi.org/10.1109/CIBCB55180.2022.9863022

Deep learning and ensemble method for optic disc and cup segmentation. / Kim, Jongwoo; Tran, Loc; Peto, Tunde et al.
2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB): proceedings. Institute of Electrical and Electronics Engineers Inc., 2022. (IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology: proceedings).

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

TY - GEN

T1 - Deep learning and ensemble method for optic disc and cup segmentation

AU - Kim, Jongwoo

AU - Tran, Loc

AU - Peto, Tunde

AU - Chew, Emily Y.

N1 - Funding Information: ACKNOWLEDGMENT This research/work was supported in part by the Lister Hill National Center for Biomedical Communications of the National Library of Medicine (NLM), National Institutes of Health. We thank the MESSIDOR program, Magrabi Eye Center in Saudi Arabia and Bin Rushed Ophthalmic center in Saudi Arabia for providing the RIGA dataset. We thank the team of Jose Ignacio Orlando for providing REFUGE dataset. In addition, we acknowledge the help of the National Eye Institute staff in providing additional fundus images (AREDS dataset) to extend the current work. Publisher Copyright: © 2022 IEEE.

PY - 2022/8/26

Y1 - 2022/8/26

N2 - Glaucoma is a chronic retinal disease that gradually damages the optic nerve. It is a leading cause of irreversible loss of vision. In ophthalmic fundus images, the cup to optic disc ratio measured around the optic nerve is a key measure used to screen for glaucomatous damages. Unfortunately, there is high subjectivity among ophthalmologists in estimating this ratio due to challenges in making reliable disc and cup measurements. To minimize this, we propose an automatic method using deep learning and ensemble method to segment the optic disc and cup. The proposed method comprises two steps. The region of interest (ROI), where optic disc is centered, is detected from a fundus image, following which the optic disc and cup are segmented from the ROI. Mask R-CNN algorithm is used to estimate the ROI, and two ensemble models based on three fully convolutional networks are used for the segmentation of optic disc and cup in parallel. The proposed method is trained and evaluated using the RIGA dataset that contains 750 fundus images and the REFUGE database containing 400 fundus images. The results demonstrate that the proposed method has a better performance compared with the current state-of-the-art algorithms. Our best segmentation results for optic disc shows 0.9303 Jaccard Index (JI) and 0.9635 Dice Coefficient (DC). The best segmentation results for cup shows 0.8096 JI and 0.8915 DC. The average cup to optic disc ratio error shows 0.0429.

AB - Glaucoma is a chronic retinal disease that gradually damages the optic nerve. It is a leading cause of irreversible loss of vision. In ophthalmic fundus images, the cup to optic disc ratio measured around the optic nerve is a key measure used to screen for glaucomatous damages. Unfortunately, there is high subjectivity among ophthalmologists in estimating this ratio due to challenges in making reliable disc and cup measurements. To minimize this, we propose an automatic method using deep learning and ensemble method to segment the optic disc and cup. The proposed method comprises two steps. The region of interest (ROI), where optic disc is centered, is detected from a fundus image, following which the optic disc and cup are segmented from the ROI. Mask R-CNN algorithm is used to estimate the ROI, and two ensemble models based on three fully convolutional networks are used for the segmentation of optic disc and cup in parallel. The proposed method is trained and evaluated using the RIGA dataset that contains 750 fundus images and the REFUGE database containing 400 fundus images. The results demonstrate that the proposed method has a better performance compared with the current state-of-the-art algorithms. Our best segmentation results for optic disc shows 0.9303 Jaccard Index (JI) and 0.9635 Dice Coefficient (DC). The best segmentation results for cup shows 0.8096 JI and 0.8915 DC. The average cup to optic disc ratio error shows 0.0429.

KW - Cup

KW - Deep Learning

KW - Ensemble Method

KW - Fully Convolutional Network (FCN)

KW - Fundus Image

KW - Glaucoma

KW - Optic Disc

U2 - 10.1109/CIBCB55180.2022.9863022

DO - 10.1109/CIBCB55180.2022.9863022

M3 - Conference contribution

AN - SCOPUS:85137905611

SN - 9781665484633

T3 - IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology: proceedings

BT - 2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB): proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology, CIBCB 2022

Y2 - 15 August 2022 through 17 August 2022

ER -

Kim J, Tran L, Peto T, Chew EY. Deep learning and ensemble method for optic disc and cup segmentation. In 2022 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB): proceedings. Institute of Electrical and Electronics Engineers Inc. 2022. (IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology: proceedings). doi: 10.1109/CIBCB55180.2022.9863022

Deep learning and ensemble method for optic disc and cup segmentation

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this