Glaucoma is the leading cause of irreversible blindness and disability worldwide. Nevertheless, the majority of patients do not know they have the disease and detection of glaucoma progression using standard technology remains a challenge in clinical practice. Artificial intelligence (AI) is an expanding field that offers the potential to improve diagnosis and screening for glaucoma with minimal reliance on human input. Deep learning (DL) algorithms have risen to the forefront of AI by providing nearly human-level performance, at times exceeding the performance of humans for detection of glaucoma on structural and functional tests. A succinct summary of present studies and challenges to be addressed in this field is needed. Following PRISMA guidelines, we conducted a systematic review of studies that applied DL methods for detection of glaucoma using color fundus photographs, optical coherence tomography (OCT), or standard automated perimetry (SAP). In this review article we describe recent advances in DL as applied to the diagnosis of glaucoma and glaucoma progression for application in screening and clinical settings, as well as the challenges that remain when applying this novel technique in glaucoma.
FlaxmanSRBourneRRAResnikoffS, et al. Global causes of blindness and distance vision impairment 1990–2020: a systematic review and meta-analysis. Lancet Glob Health2017; 5: e1221–e1234.
2.
ThamYCLiXWongTY, et al. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmology2014; 121: 2081–2090.
3.
PetersDBengtssonBHeijlA.Factors associated with lifetime risk of open-angle glaucoma blindness. Acta Ophthalmol2014; 92: 421–425.
4.
WeinrebRNAungTMedeirosFA.The pathophysiology and treatment of glaucoma: a review. J Am Med Assoc2014; 311: 1901–1911.
5.
HarwerthRSCarter-DawsonLShenF, et al. Ganglion cell losses underlying visual field defects from experimental glaucoma. Invest Ophthalmol Visual Sci1999; 40: 2242–2250.
6.
HarwerthRSCarter-DawsonLSmithEL, et al. Neural losses correlated with visual losses in clinical perimetry. Invest Ophthalmol Visual Sci2004; 45: 3152–3160.
7.
BudenzDLBartonKWhiteside-de VosJ, et al. Prevalence of glaucoma in an urban West African population: the Tema Eye Survey. JAMA Ophthalmol2013; 131: 651–658.
8.
HennisAWuS-YNemesureB, et al. Awareness of incident open-angle glaucoma in a population study: the Barbados Eye Studies. Ophthalmology2007; 114: 1816–1821.
9.
BolandMVGuptaPKoF, et al. Evaluation of frequency-doubling technology perimetry as a means of screening for glaucoma and other eye diseases using the National Health and Nutrition Examination Survey. JAMA Ophthalmol2016; 134: 57–62.
10.
KatzJTielschJMQuigleyHA, et al. Automated suprathreshold screening for glaucoma: the Baltimore Eye Survey. Invest Ophthalmol Visual Sci1993; 34: 3271–3277.
11.
ErvinAMBolandMVMyrowitzEH, et al. AHRQ comparative effectiveness reviews. Screening for glaucoma: comparative effectiveness. Rockville, MD: Agency for Healthcare Research and Quality (US), 2012.
12.
QuigleyHAWestSKRodriguezJ, et al. The prevalence of glaucoma in a population-based study of Hispanic subjects: Proyecto VER. Arch Ophthalmol2001; 119: 1819–1826.
13.
HametPTremblayJ.Artificial intelligence in medicine. Metab Clin Exp2017; 69: S36–S40.
14.
HuynhBQLiHGigerML.Digital mammographic tumor classification using transfer learning from deep convolutional neural networks. J Med Imaging2016; 3: 034501.
15.
KrittanawongCZhangHWangZ, et al. Artificial intelligence in precision cardiovascular medicine. J Am Coll Cardiol2017; 69: 2657–2664 [Review].
16.
LitjensGSánchezCITimofeevaN, et al. Deep learning as a tool for increased accuracy and efficiency of histopathological diagnosis. Sci Rep2016; 6: 26286.
17.
SamuelAL.Some studies in machine learning using the game of checkers. IBM J Res Dev2000; 44: 206–226.
18.
HogartyDTMackeyDAHewittAW.Current state and future prospects of artificial intelligence in ophthalmology: a review. Clin Exp Ophthalmol2019; 47: 128–139.
19.
BaxterSLMarksCKuoTT, et al. Machine learning-based predictive modeling of surgical intervention in glaucoma using systemic data from electronic health records. Am J Ophthalmol2019; 208: 30–40.
20.
GrewalPSOloumiFRubinU, et al. Deep learning in ophthalmology: a review. Can J Ophthalmol2018; 53: 309–313.
LeCunYBengioY.Convolutional networks for images, speech, and time series. In: ArbibMA (ed.) The handbook of brain theory and neural networks, vol. 3361. Cambridge, MA: MIT Press, 1995.
23.
HubelDHWieselTN.Receptive fields, binocular interaction and functional architecture in the cat’s visual cortex. J Physiol1962; 160: 106–154.
24.
FukushimaK.Neocognitron: a self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position. Biol Cybern1980; 36: 193–202.
25.
OrlandoJIFuHBarbossa BredaJ, et al. REFUGE Challenge: a unified framework for evaluating automated methods for glaucoma assessment from fundus photographs. Med Image Anal2019; 59: 101570.
26.
MillerSEThapaSRobinAL, et al. Glaucoma screening in Nepal: cup-to-disc estimate with standard mydriatic fundus camera compared to portable nonmydriatic camera. Am J Ophthalmol2017; 182: 99–106.
27.
AhnJMKimSAhnKS, et al. A deep learning model for the detection of both advanced and early glaucoma using fundus photography. PLoS One2018; 13: 8.
28.
ChaiYDLiuHYXuJ.Glaucoma diagnosis based on both hidden features and domain knowledge through deep learning models. Knowledge-Based Syst2018; 161: 147–156.
29.
ChristopherMBeighithABowdC, et al. Performance of deep learning architectures and transfer learning for detecting glaucomatous optic neuropathy in fundus photographs. Sci Rep2018; 8: 13.
30.
LiuSGrahamSLSchulzA, et al. A deep learning-based algorithm identifies glaucomatous discs using monoscopic fundus photographs. Ophthalmol Glaucoma2018; 1: 15–22.
31.
AsaokaRTanitoMShibataN, et al. Validation of a deep learning model to screen for glaucoma using images from different fundus cameras and data augmentation. Ophthalmol Glaucoma2019; 2: 224–231.
32.
BajwaMNMalikMISiddiquiSA, et al. Two-stage framework for optic disc localization and glaucoma classification in retinal fundus images using deep learning. BMC Med Inform Decis Mak2019; 19: 136.
33.
HemelingsRElenBBarbosa-BredaJ, et al. Accurate prediction of glaucoma from colour fundus images with a convolutional neural network that relies on active and transfer learning. Acta Ophthalmol2020; 98: e94–e100.
34.
KimMHanJCHyunSH, et al. Medinoid: computer-aided diagnosis and localization of glaucoma using deep learning. Appl Sci-Basel2019; 9: 19.
35.
LiaoWZouBZhaoR, et al. Clinical interpretable deep learning model for glaucoma diagnosis. IEEE J Biomed Health Inform2019; 24(5): 1405–1412.
36.
LiuHLiLWormstoneIM, et al. Development and validation of a deep learning system to detect glaucomatous optic neuropathy using fundus photographs. JAMA Ophthalmol2019; 137(12): 1353–1360.
37.
PheneSDunnRCHammelN, et al. Deep learning and glaucoma specialists: the relative importance of optic disc features to predict glaucoma referral in fundus photographs. Ophthalmology2019; 126(12): 1627–1639.
38.
RogersTWJaccardNCarbonaroF, et al. Evaluation of an AI system for the automated detection of glaucoma from stereoscopic optic disc photographs: the European Optic Disc Assessment Study. Eye2019; 33: 1791–1797.
39.
ShibataNTanitoMMitsuhashiK, et al. Development of a deep residual learning algorithm to screen for glaucoma from fundus photography. Sci Rep2018; 8: 9.
40.
TingDSWCheungCYLimG, et al. Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes. J Am Med Assoc2017; 318: 2211–2223.
41.
LiZHeYKeelS, et al. Efficacy of a deep learning system for detecting glaucomatous optic neuropathy based on color fundus photographs. Ophthalmology2018; 125: 1199–1206.
42.
LiFYanLWangY, et al. Deep learning-based automated detection of glaucomatous optic neuropathy on color fundus photographs. Graefe’s Arch Clin Exp Ophthalmol2020; 258: 851–867.
43.
ZhaoRChenXLiuX, et al. Direct cup-to-disc ratio estimation for glaucoma screening via semi-supervised learning. IEEE J Biomed Health Inform2020; 24: 1104–1113.
44.
Al-AswadLKapoorRChuC, et al. Evaluation of the pegasus deep learning system for identifying glaucomatous optic neuropathy based on color fundus photographs. Invest Ophthalmol Visual Sci2019; 60: 1463 [Conference abstract].
45.
OwsleyCMcGwinGLeeDJ, et al. Diabetes eye screening in urban settings serving minority populations: detection of diabetic retinopathy and other ocular findings using telemedicine. JAMA Ophthalmol2015; 133: 174–181.
46.
HuangOSTayWTOngPG, et al. Prevalence and determinants of undiagnosed diabetic retinopathy and vision-threatening retinopathy in a multiethnic Asian cohort: the Singapore Epidemiology of Eye Diseases (SEED) study. Br J Ophthalmol2015; 99: 1614–1621.
47.
ChuaJBaskaranMOngPG, et al. Prevalence, risk factors, and visual features of undiagnosed glaucoma: the singapore epidemiology of eye diseases study. JAMA Ophthalmol2015; 133: 938–946.
48.
RenSHeKGirshickR, et al. Faster r-cnn: towards real-time object detection with region proposal networks. IEEE transactions on pattern analysis and machine intelligence2017, pp. 1137–1149.
49.
XiangyuCYanwuXDamon Wing KeeW, et al. Glaucoma detection based on deep convolutional neural network. Conf Proc IEEE Eng Med Biol Soc2015; 2015: 715–718.
50.
ChenXXuYYanS, et al. Automatic feature learning for glaucoma detection based on deep learning. In: NavabNHorneggerJWellsW, et al. (eds) International conference on medical image computing and computer-assisted intervention, Munich, Germany, 5–9 October 2015, pp. 669–677. Cham: Springer, 2015.
51.
LiAChengJWongDWK, et al. Integrating holistic and local deep features for glaucoma classification. In: 2016 38th annual international conference of the IEEE engineering in medicine and biology society (EMBC), Orlando, FL, USA, 16–20 August 2016, pp. 1328–1331. Piscataway, NJ: IEEE.
52.
XuYLinSWongDW, et al. Efficient reconstruction-based optic cup localization for glaucoma screening. In: Medical image computing and computer-assisted intervention: MICCAI international conference on medical image computing and computer-assisted intervention, vol. 16, Nagoya, Japan, 22–26 September 2013, pp. 445–452. New York, NY: Springer.
53.
FuHChengJXuY, et al. Joint optic disc and cup segmentation based on multi-label deep network and polar transformation. IEEE Trans Med Imaging2018; 37: 1597–1605.
54.
AcharyaURNgEEugeneLWJ, et al. Decision support system for the glaucoma using Gabor transformation. Biomed Signal Process Control2015; 15: 18–26.
55.
DuaSAcharyaURChowriappaP, et al. Wavelet-based energy features for glaucomatous image classification. IEEE transactions on information technology in biomedicine2011; 16: 80–87.
56.
BockRMeierJNyúlLG, et al. Glaucoma risk index: automated glaucoma detection from color fundus images. Med Image Anal2010; 14: 471–481.
57.
ChengJLiuJTaoD, et al. Superpixel classification based optic cup segmentation. In: Medical image computing and computer-assisted intervention: MICCAI international conference on medical image computing and computer-assisted intervention, vol. 16, Nagoya, Japan, 22–26 September 2013, pp. 421–428. New York, NY: Springer.
58.
ZouBChenCZhuC, et al. Classified optic disc localization algorithm based on verification model. Comput Graph2018; 70: 281–287.
59.
ZhaoRChenZDuanX, et al. Automatic detection of glaucoma based on aggregated multi-channel features. J Comput Aided Des Comput Graph2017; 29: 998–1006.
60.
MedeirosFAJammalAAThompsonAC.From machine to machine: an OCT-trained deep learning algorithm for objective quantification of glaucomatous damage in fundus photographs. Ophthalmology2019; 126: 513–521.
61.
HadwinSERedmondTGarway-HeathDF, et al. Assessment of optic disc photographs for glaucoma by UK optometrists: the Moorfields Optic Disc Assessment Study (MODAS). Ophthalmic Physiol Opt2013; 33: 618–624.
62.
ReusNJLemijHGGarway-HeathDF, et al. Clinical assessment of stereoscopic optic disc photographs for glaucoma: the European Optic Disc Assessment Trial. Ophthalmology2010; 117: 717–723.
63.
ThompsonACJammalAAMedeirosFA.A deep learning algorithm to quantify neuroretinal rim loss from optic disc photographs. Am J Ophthalmol2019; 201: 9–18.
64.
GuoYZouBChenZ, et al. Optic cup segmentation using large pixel patch based CNNs. In: ChenXGarvinMKLiuJ, et al. (eds) Ophthalmic medical image analysis third international workshop, Athens, Greece, 21October2016, pp. 129–136. Iowa, IA: The University of Iowa.
65.
YinFLiuJWongDWK, et al. Automated segmentation of optic disc and optic cup in fundus images for glaucoma diagnosis. In: 2012 25th IEEE international symposium on computer-based medical systems (CBMS), Rome, Italy, 20–22 June 2012, pp. 1–6. New York, NY: IEEE.
66.
MaryMCVSRajsinghEBJacobJKK, et al. An empirical study on optic disc segmentation using an active contour model. Biomed Signal Process Control2015; 18: 19–29.
67.
ChengJYinFWongDWK, et al. Sparse dissimilarity-constrained coding for glaucoma screening. IEEE Trans Biomed Eng2015; 62: 1395–1403.
68.
JoshiGDSivaswamyJKaranK, et al. Optic disk and cup boundary detection using regional information. In: 2010 IEEE international symposium on biomedical imaging: from nano to macro, Rotterdam, Netherlands, 14–17 April 2010, pp. 948–951. New York, NY: IEEE.
69.
JoshiGDSivaswamyJKrishnadasSR.Optic disk and cup segmentation from monocular color retinal images for glaucoma assessment. IEEE Trans Med Imaging2011; 30: 1192–1205.
70.
LinKLuJChenC-S, et al. Learning compact binary descriptors with unsupervised deep neural networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition, Las Vegas, Nevada, USA, 27–30 June 2016, pp. 1183–1192. New York, NY: IEEE.
71.
WeissKKhoshgoftaarTMWangD.A survey of transfer learning. J Big Data2016; 3: 9.
72.
WongSCGattAStamatescuV, et al. Understanding data augmentation for classification: when to warp? In: 2016 International conference on digital image computing: techniques and applications (DICTA 2016), Gold Coast, QLD, Australia, 30 November to 2 December 2016. Piscataway, NJ: IEEE.
73.
HawkinsDM.The problem of overfitting. J Chem Inf Comput Sci2004; 44: 1–12.
74.
RosmanMZhengYWongW, et al. Singapore Malay eye study: rationale and methodology of 6-year follow-up study (SiMES-2). Clin Exp Ophthalmol2012; 40: 557–568.
75.
LiFWangZQuG, et al. Automatic differentiation of glaucoma visual field from non-glaucoma visual filed using deep convolutional neural network. BMC Med Imaging2018; 18: 35.
76.
VessaniRMMoritzRBatisL, et al. Comparison of quantitative imaging devices and subjective optic nerve head assessment by general ophthalmologists to differentiate normal from glaucomatous eyes. J Glaucoma2009; 18: 253–261.
77.
AbramsLSScottIUSpaethGL, et al. Agreement among optometrists, ophthalmologists, and residents in evaluating the optic disc for glaucoma. Ophthalmology1994; 101: 1662–1667.
78.
JampelHDSinghKLinSC, et al. Assessment of visual function in glaucoma: a report by the American Academy of Ophthalmology. Ophthalmology2011; 118: 986–1002.
79.
ChanHHOngDNKongYXG, et al. Glaucomatous optic neuropathy evaluation (GONE) project: the effect of monoscopic versus stereoscopic viewing conditions on optic nerve evaluation. Am J Ophthalmol2014; 157: 936–944.e931.
80.
KumarSGiubilatoAMorganW, et al. Glaucoma screening: analysis of conventional and telemedicine-friendly devices. Clin Exp Ophthalmol2007; 35: 237–243.
81.
LichterP.Variability of expert observers in evaluating the optic disc. Trans Am Ophthalmol Soc1976; 74: 532.
82.
MarcusDMBrooksSEUlrichLD, et al. Telemedicine diagnosis of eye disorders by direct ophthalmoscopy. A pilot study. Ophthalmology1998; 105: 1907–1914.
83.
WeinrebRNBowdCMoghimiS, et al. Ophthalmic diagnostic imaging: glaucoma. In: BilleJF (ed.) High resolution imaging in microscopy and ophthalmology. Berlin: Springer, 2019, pp. 107–134.
84.
LeungCK-sCheungCY-lWeinrebRN, et al. Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: a variability and diagnostic performance study. Ophthalmology2009; 116: 1257–1263.e1252.
85.
TathamAJMedeirosFA.Detecting structural progression in glaucoma with optical coherence tomography. Ophthalmology2017; 124: S57–S65.
86.
HuaZFangQShaX, et al. Role of retinal nerve fiber layer thickness and optic disk measurement by OCT on early diagnosis of glaucoma. Eye Sci2015; 30: 7–12.
87.
KuangTMZhangCZangwillLM, et al. Estimating lead time gained by optical coherence tomography in detecting glaucoma before development of visual field defects. Ophthalmology2015; 122: 2002–2009.
88.
MedeirosFAZangwillLMBowdC, et al. Evaluation of retinal nerve fiber layer, optic nerve head, and macular thickness measurements for glaucoma detection using optical coherence tomography. Am J Ophthalmol2005; 139: 44–55.
89.
RohKHJeoungJWParkKH, et al. Long-term reproducibility of cirrus HD optical coherence tomography deviation map in clinically stable glaucomatous eyes. Ophthalmology2013; 120: 969–977.
90.
DongZMWollsteinGSchumanJS.Clinical utility of optical coherence tomography in glaucoma. Invest Ophthalmol Visual Sci2016; 57: OCT556–OCT567.
91.
RussakoffDBMannilSSOakleyJD, et al. A 3D deep learning system for detecting referable glaucoma using full OCT macular cube scans. Invest Ophthalmol Visual Sci2019; 60: 1474 [Conference abstract].
92.
ZhengCXieXHuangL, et al. Detecting glaucoma based on spectral domain optical coherence tomography imaging of peripapillary retinal nerve fiber layer: a comparison study between hand-crafted features and deep learning model. Graefe’s Arch Clin Exp Ophthalmol2020; 258: 577–585.
93.
ThompsonACJammalAABerchuckSI, et al. Assessment of a segmentation-free deep learning algorithm for diagnosing glaucoma from optical coherence tomography scans. JAMA Ophthalmol2020; 138(4): 333–339.
94.
AsraniSEssaidLAlderBD, et al. Artifacts in spectral-domain optical coherence tomography measurements in glaucoma. JAMA Ophthalmol2014; 132: 396–402.
95.
MansbergerSLMendaSAFortuneBA, et al. Automated segmentation errors when using optical coherence tomography to measure retinal nerve fiber layer thickness in glaucoma. Am J Ophthalmol2017; 174: 1–8.
96.
LiuYSimavliHQueCJ, et al. Patient characteristics associated with artifacts in Spectralis optical coherence tomography imaging of the retinal nerve fiber layer in glaucoma. Am J Ophthalmol2015; 159: 565–576.e562.
97.
MikiAKumoiMUsuiS, et al. Prevalence and associated factors of segmentation errors in the peripapillary retinal nerve fiber layer and macular ganglion cell complex in spectral-domain optical coherence tomography images. J Glaucoma2017; 26: 995–1000.
98.
HardinJSTaibbiGNelsonSC, et al. Factors affecting Cirrus-HD OCT optic disc scan quality: a review with case examples. J Ophthalmol2015; 2015(5): 1–16.
99.
ChongGTLeeRK.Glaucoma versus red disease: imaging and glaucoma diagnosis. Curr Opin Ophthalmol2012; 23: 79–88.
100.
AsaokaRMurataHHirasawaK, et al. Using deep learning and transfer learning to accurately diagnose early-onset glaucoma from macular optical coherence tomography images. Am J Ophthalmol2019; 198: 136–145.
101.
AsaokaRHirasawaKIwaseA, et al. Validating the usefulness of the “random forests” classifier to diagnose early glaucoma with optical coherence tomography. Am J Ophthalmol2017; 174: 95–103.
102.
MaetschkeSAntonyBIshikawaH, et al. A feature agnostic approach for glaucoma detection in OCT volumes. PLoS One2019; 14: e0219126.
103.
RussakoffDBMannilSSOakleyJD, et al. A 3D deep learning system for detecting referable glaucoma using full OCT macular cube scans. Transl Vision Sci Technol2020; 9(2):12.
104.
AlencarLMMedeirosFA.The role of standard automated perimetry and newer functional methods for glaucoma diagnosis and follow-up. Indian J Ophthalmol2011; 59: S53.
105.
KucurSSHolloGSznitmanR.A deep learning approach to automatic detection of early glaucoma from visual fields. PLoS One2018; 13: e0206081.
106.
AsaokaRMurataHIwaseA, et al. Detecting preperimetric glaucoma with standard automated perimetry using a deep learning classifier. Ophthalmology2016; 123: 1974–1980.
107.
The Advanced Glaucoma Intervention Study (AGIS): 14. Distinguishing progression of glaucoma from visual field fluctuations. Ophthalmology2004; 111: 2109–2116.e2104.
108.
BrusiniPFilacordaS.Enhanced Glaucoma Staging System (GSS 2) for classifying functional damage in glaucoma. J Glaucoma2006; 15: 40–46.
109.
ViswanathanACrabbDMcNaughtA, et al. Interobserver agreement on visual field progression in glaucoma: a comparison of methods. Br J Ophthalmol2003; 87: 726–730.
110.
TannaAPBandiJRBudenzDL, et al. Interobserver agreement and intraobserver reproducibility of the subjective determination of glaucomatous visual field progression. Ophthalmology2011; 118: 60–65.
111.
WernerEBBishopKIKoelleJ, et al. A comparison of experienced clinical observers and statistical tests in detection of progressive visual field loss in glaucoma using automated perimetry. Arch Ophthalmol1988; 106: 619–623.
112.
SpryPGJohnsonCA.Identification of progressive glaucomatous visual field loss. Surv Ophthalmol2002; 47: 158–173.
113.
Arnalich-MontielFCasas-LleraPMuñoz-NegreteFJ, et al. Performance of glaucoma progression analysis software in a glaucoma population. Graefe’s Arch Clin Exp Ophthalmol2009; 247: 391.
114.
ThompsonACJammalAAMedeirosFA.Performance of the rule of 5 for detecting glaucoma progression between visits with OCT. Ophthalmol Glaucoma2019; 2: 319–326.
115.
ZhangXDastiridouAFrancisBA, et al. Comparison of glaucoma progression detection by optical coherence tomography and visual field. Am J Ophthalmol2017; 184: 63–74.
116.
UrataCNMariottoniEBJammalAA, et al. Comparison of short-and long-term variability in standard perimetry and spectral domain optical coherence tomography in glaucoma. Am J Ophthalmol2020; 210: 19–25.
117.
WenJCLeeCSKeanePA, et al. Forecasting future Humphrey visual fields using deep learning. PLoS One2019; 14(4): e0214875.
118.
ParkKKimJLeeJ.Visual field prediction using recurrent neural network. Sci Rep2019; 9: 1–12.
119.
BerchuckSIMukherjeeSMedeirosFA.Estimating rates of progression and predicting future visual fields in glaucoma using a deep variational autoencoder. Sci Rep2019; 9: 1–12.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
