A pixel processing approach for retinal vessel extraction using modified Gabor functions

Sameena Pathan, P. C. Siddalingaswamy, K. Gopalakrishna Prabhu

Research output: Contribution to journalArticle

Abstract

Computerized image analysis methods for retinal imaging are primarily of great interest and benefit as it provides significant information about the retinal vessels. Retinal image analysis techniques can be of pertinence for ophthalmologists and a stand-alone warning implement for determining the retinal disorders. This requires dedicated image processing algorithms to provide mathematical description about the region of interest. This paper presents an automated pixel processing-based retinal vessel extraction algorithm using modified Gabor functions and morphological operators. Color normalization is performed to make the algorithm adaptable to intra- and inter-image variabilities. Furthermore, the enhanced retinal vessels are subjected to automatic thresholding for vessel pixel classification. The proposed method is tested on a set of retinal images collected from the DRIVE database and subjected to robust performance analysis to evaluate the efficacy. The proposed algorithm achieved an average accuracy of 97.22%, sensitivity of 85.12% and specificity of 98.57%, which is comparably preferable to the well-known algorithms.

Original languageEnglish
JournalProgress in Artificial Intelligence
Volume7
Issue number1
DOIs
Publication statusPublished - 01-03-2018

Fingerprint

Pixels
Processing
Image analysis
Mathematical operators
Image processing
Color
Imaging techniques

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence

Cite this

@article{75e8e7e2434d4a20b649ec177e5d4218,
title = "A pixel processing approach for retinal vessel extraction using modified Gabor functions",
abstract = "Computerized image analysis methods for retinal imaging are primarily of great interest and benefit as it provides significant information about the retinal vessels. Retinal image analysis techniques can be of pertinence for ophthalmologists and a stand-alone warning implement for determining the retinal disorders. This requires dedicated image processing algorithms to provide mathematical description about the region of interest. This paper presents an automated pixel processing-based retinal vessel extraction algorithm using modified Gabor functions and morphological operators. Color normalization is performed to make the algorithm adaptable to intra- and inter-image variabilities. Furthermore, the enhanced retinal vessels are subjected to automatic thresholding for vessel pixel classification. The proposed method is tested on a set of retinal images collected from the DRIVE database and subjected to robust performance analysis to evaluate the efficacy. The proposed algorithm achieved an average accuracy of 97.22{\%}, sensitivity of 85.12{\%} and specificity of 98.57{\%}, which is comparably preferable to the well-known algorithms.",
author = "Sameena Pathan and Siddalingaswamy, {P. C.} and Prabhu, {K. Gopalakrishna}",
year = "2018",
month = "3",
day = "1",
doi = "10.1007/s13748-017-0134-4",
language = "English",
volume = "7",
journal = "Progress in Artificial Intelligence",
issn = "2192-6352",
publisher = "Springer Nature",
number = "1",

}

A pixel processing approach for retinal vessel extraction using modified Gabor functions. / Pathan, Sameena; Siddalingaswamy, P. C.; Prabhu, K. Gopalakrishna.

In: Progress in Artificial Intelligence, Vol. 7, No. 1, 01.03.2018.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A pixel processing approach for retinal vessel extraction using modified Gabor functions

AU - Pathan, Sameena

AU - Siddalingaswamy, P. C.

AU - Prabhu, K. Gopalakrishna

PY - 2018/3/1

Y1 - 2018/3/1

N2 - Computerized image analysis methods for retinal imaging are primarily of great interest and benefit as it provides significant information about the retinal vessels. Retinal image analysis techniques can be of pertinence for ophthalmologists and a stand-alone warning implement for determining the retinal disorders. This requires dedicated image processing algorithms to provide mathematical description about the region of interest. This paper presents an automated pixel processing-based retinal vessel extraction algorithm using modified Gabor functions and morphological operators. Color normalization is performed to make the algorithm adaptable to intra- and inter-image variabilities. Furthermore, the enhanced retinal vessels are subjected to automatic thresholding for vessel pixel classification. The proposed method is tested on a set of retinal images collected from the DRIVE database and subjected to robust performance analysis to evaluate the efficacy. The proposed algorithm achieved an average accuracy of 97.22%, sensitivity of 85.12% and specificity of 98.57%, which is comparably preferable to the well-known algorithms.

AB - Computerized image analysis methods for retinal imaging are primarily of great interest and benefit as it provides significant information about the retinal vessels. Retinal image analysis techniques can be of pertinence for ophthalmologists and a stand-alone warning implement for determining the retinal disorders. This requires dedicated image processing algorithms to provide mathematical description about the region of interest. This paper presents an automated pixel processing-based retinal vessel extraction algorithm using modified Gabor functions and morphological operators. Color normalization is performed to make the algorithm adaptable to intra- and inter-image variabilities. Furthermore, the enhanced retinal vessels are subjected to automatic thresholding for vessel pixel classification. The proposed method is tested on a set of retinal images collected from the DRIVE database and subjected to robust performance analysis to evaluate the efficacy. The proposed algorithm achieved an average accuracy of 97.22%, sensitivity of 85.12% and specificity of 98.57%, which is comparably preferable to the well-known algorithms.

UR - http://www.scopus.com/inward/record.url?scp=85056126467&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85056126467&partnerID=8YFLogxK

U2 - 10.1007/s13748-017-0134-4

DO - 10.1007/s13748-017-0134-4

M3 - Article

VL - 7

JO - Progress in Artificial Intelligence

JF - Progress in Artificial Intelligence

SN - 2192-6352

IS - 1

ER -