Share Email Print

Proceedings Paper

Glaucoma detection using statistical features: comparative study in RGB, HSV and CIEL*a*b* color models
Author(s): Lamiaa Abdel-Hamid
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Glaucoma is among the leading causes of irreversible blindness worldwide. Early detection and proper treatment can help delay disease progression and in turn associated visual impairments. Glaucoma is commonly linked with elevated eye pressure leading to damaging of the optic nerve. Retinal images have been successfully utilized for glaucoma diagnosis by extracting relevant features related to the optic disc, most commonly from the RGB color model. In this study, the relevance of statistical features computed from the detected optic disc region are compared for the RGB, HSV, and CIEL* a * b* color models. Feature computations required less than 0.2 seconds per image for the complete feature vector. Correlation feature selection was employed to determine the most relevant channels and associated features in each color model. Analysis showed that the red, value, and b* channels of the RGB, HSV, and CIEL* a * b* models, respectively were the least useful for glaucoma detection. Furthermore, statistical features computed from the HSV model were found to result in superior results in comparison to statistical features extracted from the RGB and CIEL* a * b* models. The HSV statistical features implemented in this work were able to classify retinal images into normal or glaucomatous with accuracies of 90.9% and 92.4% along with areas under the receiver operating curve (AUC) of 0.953 and 0.944 using the support machine vector (SVM) and naïve Bayes classifiers, respectively.

Paper Details

Date Published: 6 May 2019
PDF: 9 pages
Proc. SPIE 11069, Tenth International Conference on Graphics and Image Processing (ICGIP 2018), 110692V (6 May 2019); doi: 10.1117/12.2524215
Show Author Affiliations
Lamiaa Abdel-Hamid, Misr International Univ. (Egypt)

Published in SPIE Proceedings Vol. 11069:
Tenth International Conference on Graphics and Image Processing (ICGIP 2018)
Chunming Li; Hui Yu; Zhigeng Pan; Yifei Pu, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?