Segmentation of optic disc, fovea and retinal vasculature using a single convolutional neural network

Jen Hong Tan, U. Rajendra Acharya, Sulatha V. Bhandary, Kuang Chua Chua, Sobha Sivaprasad

Research output: Contribution to journalArticle

53 Citations (Scopus)

Abstract

We have developed and trained a convolutional neural network to automatically and simultaneously segment optic disc, fovea and blood vessels. Fundus images were normalized before segmentation was performed to enforce consistency in background lighting and contrast. For every effective point in the fundus image, our algorithm extracted three channels of input from the point's neighbourhood and forwarded the response across the 7-layer network. The output layer consists of four neurons, representing background, optic disc, fovea and blood vessels. In average, our segmentation correctly classified 92.68% of the ground truths (on the testing set from Drive database). The highest accuracy achieved on a single image was 94.54%, the lowest 88.85%. A single convolutional neural network can be used not just to segment blood vessels, but also optic disc and fovea with good accuracy.

Original languageEnglish
Pages (from-to)70-79
Number of pages10
JournalJournal of Computational Science
Volume20
DOIs
Publication statusPublished - 01-05-2017

Fingerprint

Blood Vessels
Blood vessels
Optics
Segmentation
Neural Networks
Neural networks
Network layers
Neurons
Neuron
Lowest
High Accuracy
Lighting
Testing
Output
Background

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)
  • Modelling and Simulation

Cite this

Tan, Jen Hong ; Acharya, U. Rajendra ; Bhandary, Sulatha V. ; Chua, Kuang Chua ; Sivaprasad, Sobha. / Segmentation of optic disc, fovea and retinal vasculature using a single convolutional neural network. In: Journal of Computational Science. 2017 ; Vol. 20. pp. 70-79.
@article{57430cdbd8564a5886732152df7dc9cc,
title = "Segmentation of optic disc, fovea and retinal vasculature using a single convolutional neural network",
abstract = "We have developed and trained a convolutional neural network to automatically and simultaneously segment optic disc, fovea and blood vessels. Fundus images were normalized before segmentation was performed to enforce consistency in background lighting and contrast. For every effective point in the fundus image, our algorithm extracted three channels of input from the point's neighbourhood and forwarded the response across the 7-layer network. The output layer consists of four neurons, representing background, optic disc, fovea and blood vessels. In average, our segmentation correctly classified 92.68{\%} of the ground truths (on the testing set from Drive database). The highest accuracy achieved on a single image was 94.54{\%}, the lowest 88.85{\%}. A single convolutional neural network can be used not just to segment blood vessels, but also optic disc and fovea with good accuracy.",
author = "Tan, {Jen Hong} and Acharya, {U. Rajendra} and Bhandary, {Sulatha V.} and Chua, {Kuang Chua} and Sobha Sivaprasad",
year = "2017",
month = "5",
day = "1",
doi = "10.1016/j.jocs.2017.02.006",
language = "English",
volume = "20",
pages = "70--79",
journal = "Journal of Computational Science",
issn = "1877-7503",
publisher = "Elsevier",

}

Segmentation of optic disc, fovea and retinal vasculature using a single convolutional neural network. / Tan, Jen Hong; Acharya, U. Rajendra; Bhandary, Sulatha V.; Chua, Kuang Chua; Sivaprasad, Sobha.

In: Journal of Computational Science, Vol. 20, 01.05.2017, p. 70-79.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Segmentation of optic disc, fovea and retinal vasculature using a single convolutional neural network

AU - Tan, Jen Hong

AU - Acharya, U. Rajendra

AU - Bhandary, Sulatha V.

AU - Chua, Kuang Chua

AU - Sivaprasad, Sobha

PY - 2017/5/1

Y1 - 2017/5/1

N2 - We have developed and trained a convolutional neural network to automatically and simultaneously segment optic disc, fovea and blood vessels. Fundus images were normalized before segmentation was performed to enforce consistency in background lighting and contrast. For every effective point in the fundus image, our algorithm extracted three channels of input from the point's neighbourhood and forwarded the response across the 7-layer network. The output layer consists of four neurons, representing background, optic disc, fovea and blood vessels. In average, our segmentation correctly classified 92.68% of the ground truths (on the testing set from Drive database). The highest accuracy achieved on a single image was 94.54%, the lowest 88.85%. A single convolutional neural network can be used not just to segment blood vessels, but also optic disc and fovea with good accuracy.

AB - We have developed and trained a convolutional neural network to automatically and simultaneously segment optic disc, fovea and blood vessels. Fundus images were normalized before segmentation was performed to enforce consistency in background lighting and contrast. For every effective point in the fundus image, our algorithm extracted three channels of input from the point's neighbourhood and forwarded the response across the 7-layer network. The output layer consists of four neurons, representing background, optic disc, fovea and blood vessels. In average, our segmentation correctly classified 92.68% of the ground truths (on the testing set from Drive database). The highest accuracy achieved on a single image was 94.54%, the lowest 88.85%. A single convolutional neural network can be used not just to segment blood vessels, but also optic disc and fovea with good accuracy.

UR - http://www.scopus.com/inward/record.url?scp=85014389195&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85014389195&partnerID=8YFLogxK

U2 - 10.1016/j.jocs.2017.02.006

DO - 10.1016/j.jocs.2017.02.006

M3 - Article

VL - 20

SP - 70

EP - 79

JO - Journal of Computational Science

JF - Journal of Computational Science

SN - 1877-7503

ER -