Towards Accurate Segmentation of Retinal Vessels and the Optic Disc in Fundoscopic Images with Generative Adversarial Networks

Jaemin Son, Sang Jun Park, Kyu Hwan Jung

Research output: Contribution to journalArticleResearchpeer-review

1 Citation (Scopus)

Abstract

Automatic segmentation of the retinal vasculature and the optic disc is a crucial task for accurate geometric analysis and reliable automated diagnosis. In recent years, Convolutional Neural Networks (CNN) have shown outstanding performance compared to the conventional approaches in the segmentation tasks. In this paper, we experimentally measure the performance gain for Generative Adversarial Networks (GAN) framework when applied to the segmentation tasks. We show that GAN achieves statistically significant improvement in area under the receiver operating characteristic (AU-ROC) and area under the precision and recall curve (AU-PR) on two public datasets (DRIVE, STARE) by segmenting fine vessels. Also, we found a model that surpassed the current state-of-the-art method by 0.2 − 1.0% in AU-ROC and 0.8 − 1.2% in AU-PR and 0.5 − 0.7% in dice coefficient. In contrast, significant improvements were not observed in the optic disc segmentation task on DRIONS-DB, RIM-ONE (r3) and Drishti-GS datasets in AU-ROC and AU-PR.

Original languageEnglish
Pages (from-to)499-512
Number of pages14
JournalJournal of Digital Imaging
Volume32
Issue number3
DOIs
StatePublished - 15 Jun 2019

Fingerprint

Retinal Vessels
Optic Disk
ROC Curve
Optics
Reaction injection molding
Neural networks
Datasets

Keywords

  • Convolutional neural network
  • Generative adversarial networks
  • Optic disc segmentation
  • Retinal vessel segmentation

Cite this

@article{adff94683ce64e8d9a9ca2eb8633a935,
title = "Towards Accurate Segmentation of Retinal Vessels and the Optic Disc in Fundoscopic Images with Generative Adversarial Networks",
abstract = "Automatic segmentation of the retinal vasculature and the optic disc is a crucial task for accurate geometric analysis and reliable automated diagnosis. In recent years, Convolutional Neural Networks (CNN) have shown outstanding performance compared to the conventional approaches in the segmentation tasks. In this paper, we experimentally measure the performance gain for Generative Adversarial Networks (GAN) framework when applied to the segmentation tasks. We show that GAN achieves statistically significant improvement in area under the receiver operating characteristic (AU-ROC) and area under the precision and recall curve (AU-PR) on two public datasets (DRIVE, STARE) by segmenting fine vessels. Also, we found a model that surpassed the current state-of-the-art method by 0.2 − 1.0{\%} in AU-ROC and 0.8 − 1.2{\%} in AU-PR and 0.5 − 0.7{\%} in dice coefficient. In contrast, significant improvements were not observed in the optic disc segmentation task on DRIONS-DB, RIM-ONE (r3) and Drishti-GS datasets in AU-ROC and AU-PR.",
keywords = "Convolutional neural network, Generative adversarial networks, Optic disc segmentation, Retinal vessel segmentation",
author = "Jaemin Son and Park, {Sang Jun} and Jung, {Kyu Hwan}",
year = "2019",
month = "6",
day = "15",
doi = "10.1007/s10278-018-0126-3",
language = "English",
volume = "32",
pages = "499--512",
journal = "Journal of Digital Imaging",
issn = "0897-1889",
publisher = "Springer New York",
number = "3",

}

Towards Accurate Segmentation of Retinal Vessels and the Optic Disc in Fundoscopic Images with Generative Adversarial Networks. / Son, Jaemin; Park, Sang Jun; Jung, Kyu Hwan.

In: Journal of Digital Imaging, Vol. 32, No. 3, 15.06.2019, p. 499-512.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Towards Accurate Segmentation of Retinal Vessels and the Optic Disc in Fundoscopic Images with Generative Adversarial Networks

AU - Son, Jaemin

AU - Park, Sang Jun

AU - Jung, Kyu Hwan

PY - 2019/6/15

Y1 - 2019/6/15

N2 - Automatic segmentation of the retinal vasculature and the optic disc is a crucial task for accurate geometric analysis and reliable automated diagnosis. In recent years, Convolutional Neural Networks (CNN) have shown outstanding performance compared to the conventional approaches in the segmentation tasks. In this paper, we experimentally measure the performance gain for Generative Adversarial Networks (GAN) framework when applied to the segmentation tasks. We show that GAN achieves statistically significant improvement in area under the receiver operating characteristic (AU-ROC) and area under the precision and recall curve (AU-PR) on two public datasets (DRIVE, STARE) by segmenting fine vessels. Also, we found a model that surpassed the current state-of-the-art method by 0.2 − 1.0% in AU-ROC and 0.8 − 1.2% in AU-PR and 0.5 − 0.7% in dice coefficient. In contrast, significant improvements were not observed in the optic disc segmentation task on DRIONS-DB, RIM-ONE (r3) and Drishti-GS datasets in AU-ROC and AU-PR.

AB - Automatic segmentation of the retinal vasculature and the optic disc is a crucial task for accurate geometric analysis and reliable automated diagnosis. In recent years, Convolutional Neural Networks (CNN) have shown outstanding performance compared to the conventional approaches in the segmentation tasks. In this paper, we experimentally measure the performance gain for Generative Adversarial Networks (GAN) framework when applied to the segmentation tasks. We show that GAN achieves statistically significant improvement in area under the receiver operating characteristic (AU-ROC) and area under the precision and recall curve (AU-PR) on two public datasets (DRIVE, STARE) by segmenting fine vessels. Also, we found a model that surpassed the current state-of-the-art method by 0.2 − 1.0% in AU-ROC and 0.8 − 1.2% in AU-PR and 0.5 − 0.7% in dice coefficient. In contrast, significant improvements were not observed in the optic disc segmentation task on DRIONS-DB, RIM-ONE (r3) and Drishti-GS datasets in AU-ROC and AU-PR.

KW - Convolutional neural network

KW - Generative adversarial networks

KW - Optic disc segmentation

KW - Retinal vessel segmentation

UR - http://www.scopus.com/inward/record.url?scp=85054686178&partnerID=8YFLogxK

U2 - 10.1007/s10278-018-0126-3

DO - 10.1007/s10278-018-0126-3

M3 - Article

VL - 32

SP - 499

EP - 512

JO - Journal of Digital Imaging

JF - Journal of Digital Imaging

SN - 0897-1889

IS - 3

ER -