Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 Nov 16:14:3881-3890.
doi: 10.2147/OPTH.S273659. eCollection 2020.

The Appropriateness of Digital Diabetic Retinopathy Screening Images for a Computer-Aided Glaucoma Screening System

Ahmed A Almazroa  1   2   3 , Maria A Woodward  1   4 , Paula Anne Newman-Casey  1   4 , Manjool M Shah  1 , Angela R Elam  1 , Shivani S Kamat  1 , Carrie A Karvonen-Gutierrez  5 , Sarah D Wood  1 , Navasuja Kumar  1 , Sayoko E Moroi  1   4
Affiliations

The Appropriateness of Digital Diabetic Retinopathy Screening Images for a Computer-Aided Glaucoma Screening System

Ahmed A Almazroa et al. Clin Ophthalmol. .

Abstract

Purpose: The purpose of this study was to evaluate the ability to screen for glaucoma using a Food Drug Administration (FDA) Class II diagnostic digital fundus photography system used for diabetic retinopathy screening (DRS).

Methods: All research participants underwent a comprehensive eye examination as well as non-mydriatic 45°single photograph retinal imaging centered on the macula. Optic nerve images within the 45° non-mydriatic and non-stereo DRS image were evaluated by two methods: 1) grading by three glaucoma specialists, and 2) a computer-aided automated segmentation system to determine the vertical cup-to-disc ratio (VCDR). Using VCDR from clinical assessment as gold standard, VCDR results from two methods were compared to that from clinical assessment. Inter-grader agreement was assessed by computing intraclass correlation coefficient (ICC). In addition, sensitivity and specificity were calculated.

Results: Among 245 fundus photos, 166 images met quality specifications for analysis. Fifty images were not processed by the automated system due to the poor quality of the optic disc, and 29 images did not include the optic nerve head due to the patient movement during the photo acquisition. The ICC value for the VCDR between the gold standard clinical exam and the automated system was 0.41, indicating fair agreement. The ICC value between the three ophthalmologists and the gold standard was 0.51, 0.56, and 0.69, respectively, indicating fair to moderate agreement.

Discussion: Assessing the VCDR on non-mydriatic and non-stereo DRS fundus photographs by either the computer-aided automated segmentation system or by glaucoma specialists showed similar fair to moderate agreement. In summary, optic nerve assessment for glaucoma from these 45° non-mydriatic and non-stereo DRS images is not yet suitable for tele-glaucoma screening.

Keywords: automated screening system; diabetic retinopathy screening; glaucoma screening; tele-glaucoma; telemedicine.

PubMed Disclaimer

Conflict of interest statement

Maria A Woodward reports grants from National Eye Institute, during the conduct of the study. Paula Anne Newman-Casey reports grants from the National Eye Institute, Career Development Award from Research to Prevent Blindness, during the conduct of the study. Dr Manjool M Shah reports personal fees from Allergan, nothing from Glaukos, nothing from Katena, outside the submitted work. Sayoko E Moroi reports grants from University of Michigan, during the conduct of the study; grants from NIH, nothing from Allergan, Royalties unrelated to this manuscript from Wolters Kluwer, outside the submitted work. The authors declare that they have no other actual or potential conflict of interests relevant to the content of this project.

Figures

Figure 1
Figure 1
The study flowchart.
Figure 2
Figure 2
Representative digital, non-stereo optic disc image shown on computer tablet. The left image (A) is the non-stereo optic disc image from a macula-centered DRS photography. The three middle images (B–D) are the same optic disc image that show the manual annotations by the three ophthalmologists as grading 3 method described in “Materials and Methods, optic nerve assessment”. The right image (E) is the automated segmentation of the optic disc and optic cup margins, grading 4. Overall the optic disc and optic cup margins showed small variations among the three ophthalmologists’ manual markings (grading 3) and the computer automated margins (grading 4).
Figure 3
Figure 3
Histogram distribution of VCDR values from four grading assessments on the 166 optic nerve images. Within each categorical VCDR range, the order from left to right is ground truth clinical exam, three ophthalmologists graded digital photos, the same three ophthalmologists did manual annotations, and automated segmentation. The X-axis represents the VCDRs, the Y-axis represents the number of images in that category.
Figure 4
Figure 4
Bland-Altman plots to compare the ground truth (grading 1) VCDR with the estimated VCDR (grading 2) of the three ophthalmologists (A) (ophthalmologist 1), (B) (ophthalmologist 2), (C) (ophthalmologist 3) and the automated VCDR (grading 4) (D). The X-axis represents the average between the two measures ie the average VCDR between grading method 1and the specified grader, the Y-axis represents the difference between the ground truth and the specified grader. The mean difference is represented by the middle line (red line). Green lines indicate the upper and lower limits of agreement.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Goh JK, Cheung CY, Sim SS, Tan PC, Tan GS, Wong TY. Retinal imaging techniques for diabetic retinopathy screening. J Diabetes Sci Technol. 2016;10(2):282–294. - PMC - PubMed
    1. Scanlon PH. The english national screening programme for diabetic retinopathy 2003–2016. Acta Diabetol. 2017;54(6):515–525. doi:10.1007/s00592-017-0974-1 - DOI - PMC - PubMed
    1. Gulshan V, Peng L, Coram M, et al. Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. J Am Med Assoc. 2016;316(22):2402–2410. doi:10.1001/jama.2016.17216 - DOI - PubMed
    1. FDA permits marketing of artificial intelligence-based device to detect certain diabetes-related eye problems. Food and drug administration. 2018. Available from: https://www.fda.gov/newsevents/newsroom/pressannouncements/ucm604357.htm. Accessed February26, 2019.
    1. Open-angle Glaucoma defined. National eye institute. 2018. Available from: https://nei.nih.gov/eyedata/glaucoma#5. Accessed February26, 2019.