Artificial intelligence in chorioretinal pathology through fundoscopy: a comprehensive review

Abstract

Background: Applications for artificial intelligence (AI) in ophthalmology are continually evolving. Fundoscopy is one of the oldest ocular imaging techniques but remains a mainstay in posterior segment imaging due to its prevalence, ease of use, and ongoing technological advancement. AI has been leveraged for fundoscopy to accomplish core tasks including segmentation, classification, and prediction.

Main body: In this article we provide a review of AI in fundoscopy applied to representative chorioretinal pathologies, including diabetic retinopathy and age-related macular degeneration, among others. We conclude with a discussion of future directions and current limitations.

Short conclusion: As AI evolves, it will become increasingly essential for the modern ophthalmologist to understand its applications and limitations to improve patient outcomes and continue to innovate.

Keywords: Age-related macular degeneration; Artificial intelligence; Choroid; Deep learning; Diabetic retinopathy; Fundoscopy; Fundus; Machine learning.

Fig. 1

Fundus vessel segmentation using a W-Net tested on a) DRIVE and b) LES-AV datasets. Reprinted with permission from Galdran et al. Galdran, A., Anjos, A., Dolz, J. et al. State-of-the-art retinal vessel segmentation with minimalistic models. Sci Rep 12, 6174 (2022) under Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0/legalcode)

Fig. 2

Automated segmentation of soft exudates (SE), hard exudates (EX), hemorrhage (HE), and microaneurysms (MA) using multiple U-Net architectures. Reprinted with permission from Xu et al. Xu Y, Zhou Z, Li X, Zhang N, Zhang M, Wei P. FFU-Net: Feature Fusion U-Net for Lesion Segmentation of Diabetic Retinopathy. Biomed Res Int. 2021 Jan 2;2021:6644071 under Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/legalcode)

Fig. 3

Representative heatmap and feature detection for classification of multiple retinal pathologies. Reprinted with permission from Cen et al. Cen, LP., Ji, J., Lin, JW. et al. Automatic detection of 39 fundus diseases and conditions in retinal photographs using deep neural networks. Nat Commun 12, 4828 (2021) under Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/legalcode)

Fig. 4

Misclassified images of DR in multiple stages due to poor lighting and contrast. Reprinted with permission from Shaban et al. Shaban M, Ogur Z, Mahmoud A, Switala A, Shalaby A, Abu Khalifeh H, Ghazal M, Fraiwan L, Giridharan G, Sandhu H, El-Baz AS. A convolutional neural network for the screening and staging of diabetic retinopathy. PLoS One. 2020 Jun 22;15 [6]:e0233514 under Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/legalcode)

Fig. 5

Classification of stages of AMD. Blue represents strong signs of AMD, while green represents weaker signs of AMD. Larger areas with more blue resulted in classification into a later stage. Reprinted with permission from Bhuiyan et al. Alauddin Bhuiyan, Tien Yin Wong, Daniel Shu Wei Ting, Arun Govindaiah, Eric H. Souied, R. Theodore Smith; Artificial Intelligence to Stratify Severity of Age-Related Macular Degeneration (AMD) and Predict Risk of Progression to Late AMD. Trans. Vis. Sci. Tech. 2020;9 [2]:25 under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (https://creativecommons.org/licenses/by-nc-nd/4.0/legalcode)