SCREENING GLAUCOMA IN COLOR FUNDUS IMAGES USING ARTIFICIAL INTELLIGENCE SOFWARE EYEDR
Main Article Content
Abstract
Using Darknet convolutional neural network technology, YOLO version 3 develops an artificial
intelligence system called EYEDR, that can screen glaucoma by disc imaging automatically.
Purposes: Determining the validity and reliability of the artificial intelligence algorithm EYEDR in
the screening of adult glaucoma by color images of the optic disc.
Methods: Cross-sectional study. The 2470 color images of the optic disc were selected to train
machine learning with Darknet YOLO convolutional neural network to determine the shape of the
optic disc and optic cup, thereby an artificial intelligence algorithm was developed to detect the
glaucoma and non-glaucoma optic nerve. A test dataset of 1028 disc images was used to evaluate
the value of this AI software based on the sensitivity, specificity, area under the ROC curve (AUC),
likelihood ratio, accuracy, and Cohen’s Kappa coefficient.
Results: The area under the curve (AUC) was 0.93 ± 0.01 (95% CI; 0.92 - 0.95), with the sensitivity
90.03%, specificity 95.06%, the accuracy 86.8% and the Kappa coefficient 0.86 with p<0.001.
Conclusions: EYEDR artificial intelligence software developed by CNN Darknet YOLO vesion 3
technology with high sensitivity could screen glaucoma by color fundus images in the community.
Article Details
Keywords
Glaucoma, optic disc, atrificial intelligence.
References
model for the detection of both advanced and
early glaucoma using fundus photography” PLOS
ONE, 1-8, 2018.
[2] Chakrabarty L, Gopal DJGD, Arunava CA et
al., Automated Detection of Glaucoma From
Topographic Features of the Optic Nerve Head
in Color Fundus Photographs. J Glaucoma.
25(7):590-597, 2016.151
[3] Chen X, Xu Y, Wong KW et al., “Glaucoma
detection based on deep convolutional neural
network”. IEEE Eng Med Biol Soc, pp. 715 –
718, Annual Conf, 2015.
[4] Fingeret M, Medeiros FA, Susanna R et al., Five
rules to evaluate the optic disc and retinal nerve
fiber layer for Glaucoma. Optometry.;76:661-8,
2005.
[5] Kauppi T, Kalesnykiene V, Kamarainen JK et
al., “DIARETDB1 diabetic retinopathy database
and evaluation protocol”. In: Medical image
understanding and analysis, processdings of 11th
conference; 2007.
[6] Li Z, He Y, Keel S et al., “Efficacy of a Deep
Learning System for Detecting Glaucomatous
Optic Neuropathy Based on Color Fundus
Photographs” American academy of
Ophthalmology, vol. 125, pp. 1199 – 1206, 2018.
[7] Liu H, Li L, Wormstone IM et al., “Development
and Validation of a Deep Learning System to
Detect Glaucomatous Optic Neuropathy using
Fundus Photographs” JAMA Ophthalmol. E1-
E8, 2019.
[8] Phene S, Dunn RC, Hammel N et al., “Deep
learning and Glaucoma specialists: The relative
importance of optic disc features to predict
Glaucoma referral in fundus photographs”.
Ophthalmology. 2019;1-13.
[9] Qaisar A, “Glaucoma-Deep: Detection of
Glaucoma Eye Disease on Retinal Fundus Images
using Deep Learning” International Journal of
Advanced Computer Science and Applications,
8(6): 41-45, 2017.
[10] Redmon J, Divvala S, Girshick R et al., “You
Only Look Once: Unified, Real-time Object
Detection,” in Computer vision and Pattern
Recognition 2016, 2016.
[11] Redmon J, Farhadi A, “Yolov3: An incremental
improvement”. arXiv Operational, 2018.
[12] Thanh THP, Thuy TPT, Hieu TN et al., “A RealTime Classification Of Glaucoma from Retinal
Fundus Images Using AI Technology,” 2020
International Conference on Advanced Computing
and Applications (ACOMP), 2020, pp. 114-121,
doi: 10.1109/ACOMP50827.2020.00024.
[13] Ting DSW, Pasquale LR, Peng L et al., “Artificial
intelligence and deep learning in ophthalmology”.
Br J Ophthalmol;103:167–175, 2019.