PLOS ONE

RESEARCH ARTICLE

Deep learning for detection of age-related

macular degeneration: A systematic review
and meta-analysis of diagnostic test accuracy
studies
Xiangjie Leng1, Ruijie Shi1, Yanxia Wu1,2, Shiyin Zhu1, Xingcan Cai1, Xuejing Lu ID1,2,3,4*,
Ruobing Liu5

1 Eye College, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China,
2 Department of Ophthalmology, Ineye Hospital of Chengdu University of Traditional Chinese Medicine,
Chengdu, Sichuan, China, 3 Department of Ophthalmology, Key Laboratory of Sichuan Province
Ophthalmopathy Prevention & Cure and Visual Function Protection with TCM Laboratory, Chengdu, Sichuan,
a1111111111 China, 4 Department of Ophthalmology, Retinal Image Technology and Chronic Vascular Disease
a1111111111 Prevention & Control and Collaborative Innovation Center, Chengdu, Sichuan, China, 5 Faculty of
Technology, Policy and Management, Delft University of Technology, Delft, South Holland, Netherlands
a1111111111
a1111111111 * [email protected]
a1111111111

Abstract
OPEN ACCESS

Citation: Leng X, Shi R, Wu Y, Zhu S, Cai X, Lu X,

Objective
et al. (2023) Deep learning for detection of age-
related macular degeneration: A systematic review To evaluate the diagnostic accuracy of deep learning algorithms to identify age-related mac-
and meta-analysis of diagnostic test accuracy ular degeneration and to explore factors impacting the results for future model training.
studies. PLoS ONE 18(4): e0284060. https://doi.
org/10.1371/journal.pone.0284060

Editor: Jude Hemanth, Karunya Institute of Methods

Technology and Sciences, INDIA
Diagnostic accuracy studies published in PubMed, EMBASE, the Cochrane Library, and
Received: January 8, 2023
ClinicalTrails.gov before 11 August 2022 which employed deep learning for age-related
Accepted: March 22, 2023 macular degeneration detection were identified and extracted by two independent research-
Published: April 6, 2023 ers. Sensitivity analysis, subgroup, and meta-regression were performed by Review Man-
Copyright: © 2023 Leng et al. This is an open ager 5.4.1, Meta-disc 1.4, and Stata 16.0. The risk of bias was assessed using QUADAS-2.
access article distributed under the terms of the The review was registered (PROSPERO CRD42022352753).
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
Results
author and source are credited.
The pooled sensitivity and specificity in this meta-analysis were 94% (P = 0, 95% CI 0.94–
Data Availability Statement: All relevant data are
within the manuscript and its Supporting 0.94, I2 = 99.7%) and 97% (P = 0, 95% CI 0.97–0.97, I2 = 99.6%), respectively. The pooled
Information files. positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio, and the area under
Funding: National Natural Science Foundation of the curve value were 21.77(95% CI 15.49–30.59), 0.06 (95% CI 0.04–0.09), 342.41 (95%
China, grant number 82174444. CI 210.31–557.49), and 0.9925, respectively. Meta-regression indicated that types of AMD
Competing interests: The authors have declared (P = 0.1882, RDOR = 36.03) and layers of the network (P = 0.4878, RDOR = 0.74) contrib-
that no competing interests exist. uted to the heterogeneity.

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 1 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Conclusions
Convolutional neural networks are mostly adopted deep learning algorithms in age-related
macular degeneration detection. Convolutional neural networks, especially ResNets, are
effective in detecting age-related macular degeneration with high diagnostic accuracy.
Types of age-related macular degeneration and layers of the network are the two essential
factors that impact the model training process. Proper layers of the network will make the
model more reliable. More datasets established by new diagnostic methods will be used to
train deep learning models in the future, which will benefit for fundus application screening,
long-range medical treatment, and reducing the workload of physicians.

Introduction
Age-related macular degeneration (AMD) is one of the leading causes of severe irreversible
vision impairment in developed countries [1, 2]. With the accelerated aging process of the
global population, the number of AMD patients is expected to increase to 288 million by 2040
[3], and it has become one of the key topics in the research of ophthalmic blindness
prevention.
Clinically, it is classified as dry AMD (dAMD) characterized by medium-sized drusen and
retinal pigmentary changes, and wet AMD (wAMD) characterized by neovascular and atro-
phic [4]. Fundus photography (FP) and optical coherence tomography (OCT) are the most
widely used auxiliary examinations in ophthalmology. FP is the cheapest and the most neces-
sary fundus test in AMD, which can intuitively identify lesions and diagnose AMD. OCT uses
low coherence light to scan biological tissues in cross-section and converts the acquired infor-
mation into numbers. After computer processing, it displays the pathological changes of each
layer of the retina clearly and provides quantitative diagnostic indicators. In addition, OPTOS
ultra-widefield retinal images can clearly visualize peripheral retinal lesions, and when com-
bined with angiography, it can clearly show peripheral choroidal neovascularization (CNV)
[5], and produces better pseudocolor images than conventional 45˚ FP in diagnosis [6]. AMD
first affects the retinal pigment epithelium, Bruch’s membrane, and choroidal capillaries in the
macular area. AMD can be manifested as drusen, atrophy of the outer retinal structure, CNV,
polypoid lesions, and pigment epithelial detachment in OCT images.
The rapid increase in the demand for screening and follow-up of AMD means that a large
number of human and financial resources need to be provided by the healthcare systems of
various countries. The use of deep learning (DL) model technology may be a long-term solu-
tion for screening and monitoring patients in primary eye care settings.
The DL model is a branch of machine learning, composed of neural networks that are good
at computer vision, perception, and image recognition. DL model uses multilayer nonlinear
information processing modules to extract supervised or unsupervised features from a set of
training data and make the correct prediction. In recent years, DL models have been widely
used in ophthalmology [7–9], dermatology [10], radiology [11, 12], pathology [13, 14], and
many other image-centric specialties. In ophthalmology-related research, DL models are
beginning to be widely used in the diagnosis and recognition of diseases including diabetic ret-
inopathy [15–17], AMD [15, 18–20], glaucoma [21], refractive error [22], and prematurity reti-
nopathy of prematurity [23–25].
To establish a DL system, technical network and the datasets are the most essential compo-
nents. Although not all CNN (Convolutional Neural Network) belongs to deep learning, CNN

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 2 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

is the most widely used technical network in AMD diagnostic research which can operate on
the whole images without requiring radiologists or ophthalmologists to manually contour on
images [26]. A CNN can be divided into input, hidden, and output layers. The hidden layers are
usually composed of convolutional, pooling, full connection, and normalization layers. The
core of the CNN is the convolutional layer, which transforms the input data by applying a set of
filters (also known as kernels) that act as feature detectors. A CNN learns the values of these fil-
ters’ weights on its own during the training process [27]. Activations are used after convolution.
The pooling layers can reduce the dimensionality and keep the most important information.
The output of the convolutional and pooling layers represents high-level features of the input
image. The purpose of the fully connected layer is to use these high-level features to classify the
input image categories based on the training dataset. Afterwards, backpropagation is performed
to calculate the network weights, and gradient descent is used to update all filters and parameter
values to minimize the output error [27]. This process will be repeated many times.
The datasets for AMD detection are various. Most public databases were established using
FP and OCT images. Peking University collected a structured FP database of 5,000 patients
including normal, diabetes, glaucoma, cataract, AMD, hypertension (H), myopia, and other
diseases/abnormalities in 2019. The database is named as Ocular Disease Intelligent Recogni-
tion (ODIR) [28]. iChallenge-AMD is composed of AMD and non-AMD (myopia, normal
control, etc.) FPs [29]. Srinivasan et al [30] conducted an OCT database (Duke dataset) that
was acquired from 45 patients: 15 normal patients, 15 patients with dry AMD, and 15 patients
with DME in 2014. Established by Rasti et al [31] in 2017, the Noor dataset was acquired at
Noor Eye Hospital in Tehran and is consisting of 50 normal, 48 dAMD, and 50 DME OCTs.
Regarding the Kaggle dataset [32], OCT images were selected from retrospective cohorts of
adult patients from the Shiley Eye Institute of the University of California San Diego, the Cali-
fornia Retinal Research Foundation, Medical Center Ophthalmology Associates, the Shanghai
First People’s Hospital, and Beijing Tongren Eye Center between July 1, 2013 and March 1,
2017. Kermany et al [33] established an OCT database (Mendeley dataset) that contains CNV,
DME, Drusen and normal people in 2018. Gholami et al [34] established an AMD retinal OCT
images database including 55 AMD images called OCTID (Optical Coherence Tomography
Image Database) in 2019. Besides public datasets, plenty of studies choose self-built databases
which obtained data from hospitals directly.
DL models, especially CNN, have flourished rapidly in AMD detection in recent years.
Although most of the DL models show effective diagnostic accuracy, DL specialists are still try-
ing to explore the best networks, diameters, and layers of the network for higher accuracy.
This meta-analysis summarized the DL models for AMD diagnosis and aimed to evaluate the
diagnostic accuracy of DL models and to explore the best settings for future AMD model train-
ing, which will benefit researchers interested in DL for the diagnosis of fundus disorder.

Methods
This systematic review and meta-analysis was conducted according to the Preferred Reporting
Items for Systematic Reviews and Meta-Analysis of Diagnostic Test Accuracy Studies (PRIS-
MA-DTA) [35], and the Cochrane handbook [36]. The PRISMA-DTA checklists are available
in S1 and S2 Tables. This meta-analysis was registered on PROSPERO (ID:
CRD42022352753).

Eligibility criteria
All peer-reviewed and preprint original articles that reported the sensitivity and specificity of
DL models in detecting AMD were considered. The detailed inclusion criteria were as follows:

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 3 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

(1) diagnosing AMD by DL model via various images; (2) true positive (TP), false positive
(FP), true negative (TN), and false negative (FN) could be obtained or transferred from the
study. Records without available data such as reviews, conference abstracts, letters, and replies
were excluded. There was no restriction on the year of publication, language, country, or
datasets.

Information sources, search strategy and study selection

The search engines used included PubMed, EMBASE, the Cochrane Library, Web of Science,
Scopus, ScienceDirect ClinicalTrails.gov, and World Health Organization International Clini-
cal Trial Registration Platform (WHO ICTRP), and Chinese Clinical Trail Registry (ChiCTR)
by 11 August 11, 2022.
The search strategy using medical subject headings (MeSH and Emtree) combined with
entry words for all search engines. Detailed search strategies in different search engines are
detailed in S1 File.
Endnote 20 was adopted for the study selection process. Duplicate studies were excluded by
automation tools. The titles and abstracts were independently identified for possible inclusion
by two authors (Leng X. and Shi R.). Disagreements were resolved by a third researcher (Wu
Y.). After full text selection, the reports assessed eligibility were included in this meta-analysis.

Data collection process and definitions for data extraction

The data from the included studies were extracted by an individual researcher (Cai X.) and
were rechecked by another (Zhu S.). The data we extracted included the first author and pub-
lished year, country, number of images, network layers, device, hardware, type of AMD data-
sets, total dataset size, type of images, TP, FP, FN, TN, AUC, sensitivity, and specificity.
AMD, including dAMD and wAMD, was considered as the target condition. The reference
standard was clinically proven AMD, while the DL-based diagnosis was considered the index
test.

Risk of bias and applicability

The study risk of bias assessment was conducted by two individual researchers using the QUA-
DAS-2 (Quality Assessment of Diagnostic Accuracy Studies-2) tool (Leng X., Shi R.). Parame-
ters included patient selection, index test, reference standard, flow and timing, and
applicability concerns in terms of patient selection, index test, and reference standard. Dis-
agreements were solved with consensus by the third researcher (Lu X.). Deeks’ funnel plot
mapped by Stata 16.0 was applied to assess the potential publication bias. An asymmetrical
funnel shape or a P < 0.05 means the presentation of publication bias [37].

Diagnostic accuracy measures and synthesis of results

To evaluate the diagnostic accuracy of deep learning in detecting AMD, the sensitivity, speci-
ficity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds
ratio (DOR) along with a 95% Confidence Interval (CI) were calculated separately for each
study. Random effects models were applied in the calculation of the pooled results.

Meta-analysis and additional analysis

Separate and summary results of sensitivity and specificity would be presented in a form of a
forest plot. The heterogeneity of the meta-analysis was evaluated by the Cochran Q-test and I2
[38]. I2 exceeding 25%, 50% and 75% indicate the meta-analysis with low, medium, and high

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 4 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

heterogeneity respectively [39]. Sensitivity analyses, subgroup analyses, and meta-regression

were conducted to explore the sources of heterogeneity. All meta-analyses and additional anal-
yses were performed using Metadisc 1.4 and Review manager 5.4.1.

Results
Study selection
The detailed study selection process is described in Fig 1. 1045 records were searched using
the present search strategy. 359 records remained after eliminating duplicate records and the
ineligible records marked by automation tools. 272 records were excluded by screening titles
and abstracts. 87 reports were sought for retrieval, of which 6 reports were not retrieved. 81

Fig 1. Study selection flow diagram.

https://doi.org/10.1371/journal.pone.0284060.g001

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 5 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

reports were assessed for eligibility through full text reading, and 2 conference abstracts, 39
irrelevant studies, and 22 no-available-data studies were excluded. Finally, 18 eligible studies
were extracted from the remaining articles by full text review.

Study characteristics
The detailed studies characteristics are summarized in Table 1. The 18 studies included were
reported as full-text articles which consist of 56 models and summarized data from 778052 var-
ious images. OCT images, FPs, OPTOS ultra-widefield retinal images, and OCT images com-
bined with FP images were included in 10, 5, 1, and 3 studies, respectively. All studies were
published in 2017–2022, which were conducted in China, the USA, Japan, India, Jordan, the
United Kingdom, Turkey, Russia, South Korea, Singapore, Norway and Spain. As for the vari-
ants, 5 studies used VGG, and 5 studies used ResNet. CapsNet, Darknet and other networks
such as AlexNet, DenseNet, and self-created networks were adopted in one study, respectively.
The layers of the network were divided into five classes including �10, 10–20, 20–50, 50–100,
and >100, which were adopted in 2, 10, 6, 2 and 1 studies, respectively.

Risk of bias and bias of publication

The results of the QUADAS-2 analysis are summarized in Fig 2. Generally, the risk of bias is
low for this meta-analysis. The risk of patient selection was considered “low risk” in 16 studies
and “unclear risk” in 2 studies. The risk of bias for the index test and reference standard was
“low risk” in all studies. The risk of bias for reference standard was rated “low” in 17 studies
and 1 were rated “unclear risk”. The risk of bias for flow and timing was rated “low” in 15 stud-
ies and 3 were rated “unclear risk”. Applicability concerns including patient selection, index
test, and the reference standard only existed in one “unclear risk” study and the other 17 stud-
ies were rated “low risk”.
Deeks’ funnel plot (Fig 3) was adopted to investigate the potential bias of publication by
Stata 16.0 (P = 0.375, 95%CI -292.9264 to 112.091), which indicated no obvious publication
bias existed in this meta-analysis.

Results of individual studies

In this research, various DL models were mentioned in the included studies for AMD identifi-
cation, including VGG, CapsNet, ResNet, AlexNet, DenseNet, ResNext, DPN, CliqueNet etc.
The results show that these models have high sensitivity and specificity in AMD identification,
which can meet the needs of practical clinical applications. The detailed results of individual
studies are summarized in Table 2. Alqudah et al [40] used a 15-layer CNN to classify 136,187
OCT images from Mendeley, Duke, and Self-built datasets (4 classes including AMD, CNV,
DME, and normal) for AMD identifying with a sensitivity of 100% and a specificity of 100%.
Bhatia et al [41] used VGG-16 to classify 5588 OCT images from Mendeley, Duke, Noor, and
Self-built datasets (4 classes including AMD, CNV, DME, normal) for AMD identifying with a
sensitivity of 94% and a specificity of 90%. Celebi et al [42] used CapsNet with 7 layers to clas-
sify 726 OCT images form Kaggle and self-built datasets (2 classes including AMD and nor-
mal) for AMD identifying with a sensitivity of 100% and a specificity of 99%. Dong et al [43]
used a joint CNN detector using Yolov3 to classify 208758 FP images from self-built multicen-
ter real-world data (11 classes including AMD, DR, glaucoma, pathological myopia, retinal
vein occlusion, macula hole, epiretinal macular membrane, hypertensive retinopathy, myelin-
ated fibers, retinitis pigmentosa and normal) for AMD identifying with a sensitivity of 88%
and a specificity of 98%. Gour et al [44] used VGG-16 to classify 331 FP images from ODIR
dataset (8 classes including AMD, cataract, diabetes, glaucoma, hyperattention, myopia, and

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 6 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Table 1. Characteristics of included studies.

First author & Country Variant of Layers GPU Datasets Number of Images Other images in datasets AMD
published year CNN Images
Alqudah 2019 Jordan Self-created 19 Nvidia Tesla K40 Duke, Mendeley, 135596 OCT CNV, DME, Normal All
[40] (12GB) Duke, Self-built
Bhatia 2019 [41] UK VGG 16 NA Mendeley, Noor, 5588 OCT CNV, DME, Normal All
Self-built
Celebi 2022 [42] Turkey CapsNet 7 Nvidia Tesla K40 Kaggle dataset, 726 OCT Normal All
(12GB) Self-built
Dong 2022 [43] China Darknet 53 NA Multicenter Self- 208758 FP Normal All
built
Gour 2020 [44] India VGG 16 NA ODIR 331 FP Cataract, Diabetes, Glaucoma, All
Hyperattention, Myopia, other
abnormalities, Normal
He 2022 [45] China ResNet 50 NA Duke, Mendeley 795 OCT DME, Normal All
Kadry 2021 [46] Norway VGG 16 NA iChallenge-AMD 6400 OCT, Non-AMD All
19 database, OCTID FP
AlexNet 11
ResNet 50
Lee 2017 [47] USA VGG 16 NVIDIA Pascal Self-built 101002 OCT Normal All
Titan X (12GB)
Ma 2022 [48] USA ResNet 34 Nvidia V100 (32 Self-built 73 OCT PCV Wet
GB)
Mathews 2022 India Self-created 11 NA Duke, Mendeley 75 OCT DME, Normal Dry
[49]
Matsuba 2019 Japan DCNN 7 NA Self-built 364 OPTOS2 Normal Wet
[50]
Motozawa 2019 Japan Unclear 18 GTX 1080 TI Self-built 169 OCT Normal All
[51] (11GB)
Takhchidi 2021 Russia ResNet 50 Nvidia RTX 2070 Self-built 1200 FP Normal All
[52] Max-Q (8GB)
Tan 2018 [53] Singapore Unclear 14 NA Self-built 1110 FP Normal All
Thomas 2021 India Unclear 19 Nvidia RTX2080 Duke, Mendeley, 1139 OCT Normal All
[54] (8GB) Noor, OCTID
Wang 2019 [55] China DenseNet 121 NVIDIA RTX 2080 Duke, Noor 8315 OCT DME, Normal All
ResNet 50 TI (11G)
ResNext 101
DPN 92
CliqueNet 10
Yoo 2018 [56] Korea VGG 19 NVIDIA GTX1060 Project Macula 83 OCT, Normal All
(3GB); GTX980 FP
(6GB)
Zapata 2020 [57] Spain Self-created 24 NA Optretina’s tagged 306302 OCT, GON3 All
dataset FP

NA, not applicable; OPTOS, OPTOS ultra-widefield retinal images; GON, glaucomatous optic neuropathy

https://doi.org/10.1371/journal.pone.0284060.t001

other abnormalities) for AMD identifying with a sensitivity of 6% and a specificity of 94%. He
et al [45] used ResNet-50 to classify 795 OCT images from Mendeley and Duke datasets (3
classes including AMD, DME, and normal) for AMD identifying with a sensitivity of 96% and
a specificity of 99%. Kadry et al [46] used VGG-16, VGG-19, AlexNet, and ResNet-50 to clas-
sify 3200 FP images and 3200 OCT images from iChallenge AMD database, OCTID (2 classes
including AMD and Non-AMD) resulting in sensitivity of 88%, 84%, 88%, 88% and specificity

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 7 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Fig 2. QUADAS-2 results in each study.

https://doi.org/10.1371/journal.pone.0284060.g002

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 8 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Fig 3. Deek’s funnel plot.

https://doi.org/10.1371/journal.pone.0284060.g003

of 85%, 87%, 85%, 84%, respectively. Lee et al [47] used VGG-16 to classify 101002 OCT
images from self-built dataset (2 classes including AMD and normal) for AMD identifying
with a sensitivity of 90% and a specificity of 91%. Ma et al [48] used ResNet-34 to classify 73
OCT images from self-built dataset (2 classes including AMD and polypoidal choroidal vascu-
lopathy) for AMD identifying with a sensitivity of 92% and a specificity of 90%. Mathews et al
[49] used a 11-layer lightweight CNN to classify 75 OCT images from Duke and Mendeley
datasets (3 classes including AMD, DME, and normal) for AMD identifying with a sensitivity
of 100% and a specificity of 100%. Matsuba et al [50] used a 7-layer CNN to classify 364
OPTOS ultra-widefield retinal images from self-built dataset (2 classes including AMD and
normal) for AMD identifying with a sensitivity of 100% and a specificity of 97%. Motozawa
et al [51] used an 18-layer CNN to classify 169 OCT images from self-built database (2 classes
including AMD and normal) for AMD identifying with a sensitivity of 99% and a specificity of
100%. Takhchidi et al [52] used ResNet-50 to classify 1200 FP images from self-built dataset (2
classes including AMD and normal) for AMD identifying with a sensitivity of 90% and a speci-
ficity of 86%. Tan et al [53] used a 14-layer CNN to classify 1110 FP images from self-built
dataset (2 classes including AMD and normal) for AMD identifying with a sensitivity of 96%
and a specificity of 94%. Thomas et al [54] used a 14-layer CNN to classify 1139 OCT images
from Mendeley, Duke, Noor, and OCTID datasets (2 classes including AMD and normal) for
AMD identifying with a sensitivity of 99% and a specificity of 100%. Wang et al [55] used Den-
seNet, ResNet, ResNext, DPN, and CliqueNet to classify 8315 OCT images from Duke and
Noor datasets (3 classes including AMD, DME and normal) resulting in sensitivity of 96%,
97%, 100%, 97%, 99% and specificity of 95%, 100%, 100%, 97%, 99% in dataset 1, and sensitiv-
ity of 95%, 100%, 99%, 100%, 93% and specificity of 95%, 99%, 95%, 99%, 98% in dataset 2.

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 9 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Table 2. Summary of each included studies.

Study Methodology Datasets Number of Classes Other diseases Sensitivity Specificity Limitations
images
Alqudah 15-layer CNN Duke, 136187 4 CNV, DME, Normal 100% 100%
2019 [40] Mendeley, OCT
Self-built images
Bhatia 2019 VGG-16 Duke, 5588 OCT 4 CNV, DME, Normal 94% 90% 1) Ignored bad quality pictures.
[41] Mendeley, images
Noor, Self-
built
Celebi 2022 CapsNet Kaggle 726 OCT 2 Normal 100% 99% 1)Did not study other retinal diseases;
[42] dataset, images 2)Ignored bad quality pictures and
patients who had other retinal diseases.
Dong 2022 A joint CNN Multicenter 208758 FP 11 DR, Glaucoma, 88% 98% 1)Only small number of retinitis
[43] detector using Self-built images Pathological myopia, pigmentosa.
Yolov3 Retinal vein occlusion,
Macula hole, Epiretinal
macular membrane,
Hypertensive retinopathy,
Myelinated fibers, Retinitis
pigmentosa, Normal
Gour 2020 VGG-16 ODIR 331 FP 8 Cataract, Diabetes, 6% 94% 1)The dataset contained 8 types of
[44] images Glaucoma, diseases, but with a small dataset.
Hyperattention, Myopia,
other abnormalities,
Normal
He 2022 ResNet-50 Duke, 795 OCT 3 DME, Normal 96% 99% 1)Only contained one other diseases.
[45] Mendeley images
Kadry 2021 VGG-16 iChallenge- 3200 FP 2 Non-AMD 88% 85% 1)The definition of non-AMD is not
[46] VGG-19 AMD and 3200 84% 87% clear.
database, OCT
AlexNet, OCTID images 88% 85%
ResNet-50 88% 84%
Lee 2017 VGG-16 Self-built 101002 2 Normal 90% 91% 1)Included only images from patients
[47] OCT who met the study criteria, and the
images neural network was only trained on
these images;
2) This model was trained using
images from a single academic center,
and the external generalizability is
unknown
Ma 2022 ResNet-34 Self-built 73 OCT 2 Polypoidal choroidal 92% 90% 1) Small dataset
[48] images vasculopathy
Mathews A 11-layer Duke, 10907 3 DME, Normal 100% 100% 1) This study used drusen macular
2022 [49] lightweight Mendeley OCT degeneration for AMD diagnosis;
CNN images 2)Only contain one other diseases.
Matsuba A 7-layer CNN Self-built 364 2 Normal 100% 97% 1) It is difficult to acquire precise
2019 [50] OPTOS images using OPTOS when the
images transmission of light into the eye is
impaired by an intermediate
translucent zone;
2) Most AMD patients accept
treatment which may cause diagnostic
error
3) Did not study other retinal diseases.
Motozawa An 18-layer Self-built 169 OCT 2 Normal 99% 100% 1) Excluded low quality images and
2019 [51] CNN images patients who had other concomitant
diseases;
2) Did not study other retinal diseases.
Takhchidi ResNet-50 Self-built 1200 FP 2 Normal 90% 86% 1) Did not study other retinal diseases.
2021 [52] images
(Continued )

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 10 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Table 2. (Continued)

Study Methodology Datasets Number of Classes Other diseases Sensitivity Specificity Limitations
images
Tan 2018 A 14-layer Self-built 1110 FP 2 Normal 96% 94% 1) Did not study other retinal diseases.
[53] CNN images
Thomas A 19-layer Mendeley, 1139 OCT 2 Normal 99% 100% 1) Did not study other retinal diseases.
2021 [54] CNN Duke, Noor, images
OCTID
Wang 2019 DenseNet-121 Duke, Noor 8315 OCT 3 DME, Normal 96% in 95% in 1) Only contained one other diseases.
[55] images Duke, 95% Duke, 95%
in Noor in Noor
ResNet-50 97% in 100% in
Duke, 100% Duke, 99%
in Noor in Noor
ResNext-101 100% in 100% in
Duke, 99% Duke, 95%
in Noor in Noor
DPN-92 97% in 97% in
Duke, 100% Duke, 99%
in Noor in Noor
CliqueNet-10 99% in 99% in
Duke, 93% Duke, 98%
in Noor in Noor
Yoo 2018 VGG-19 Project 83 FP and 2 Normal 84% 59% 1) Did not study other retinal diseases;
[56] Macula 83 OCT 2) Small datasets;
images
Zapata A 24-layer Optretina’s 306302 FP 2 Glaucomatous optic 83% 89% 1.No clear number of OCT or FP
2020 [57] CNN tagged dataset images neuropathy images.
and OCT
images
https://doi.org/10.1371/journal.pone.0284060.t002

Yoo et al [56] used VGG-19 to classify three types of images (OCT, FP, and OCT combined
with FP) from Project Macula (2 classes including AMD and normal) for AMD identifying
with a pooled sensitivity of 84% and a pooled sensitivity of 59%. Zapata et al [57] used a
24-layer CNN to classify 306302 FP images and OCT images from Optretina’s tagged dataset
(2 classes including AMD and glaucomatous optic neuropathy) for AMD identifying with a
sensitivity of 83% and a specificity of 89%.

Results of synthesis
The pooled sensitivity and specificity in this meta-analysis were 94% (P = 0, 95% CI 0.94–0.94,
I2 = 99.7%) and 97% (P = 0, 95% CI 0.97–0.97, I2 = 99.6%) (Fig 4, S1 Fig), respectively. The
PLR, NLR, DOR, and AUC values were 21.77(95% CI 15.49–30.59), 0.06 (95% CI 0.04–0.09),
342.41 (95% CI 210.31–557.49) and 0.9925. The SROC (Summary Receiver Operating Charac-
teristic) curves are showed in Fig 5(A).

Additional analysis
For the high heterogeneity, the additional analyses were conducted based on the results of sen-
sitivity and specificity. Sensitivity analyses were conducted to investigate the sources of hetero-
geneity, however, neither the I2 of sensitivity nor specificity significantly decreased after
excluding studies one by one. Therefore, subgroup analyses which included the type of AMD,
type of images, variant of CNN, and variants were conducted (Fig 5B–5E). Meta regression
indicated that the sources of heterogeneity were types of AMD (P = 0.1882, RDOR = 36.03)

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 11 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Fig 4. The forest plot of the pooled sensitivity and specificity.

https://doi.org/10.1371/journal.pone.0284060.g004

and layers of the network (P = 0.4878, RDOR = 0.74). All additional analyses results were sum-
marized in Table 3. Original forest plots were available in S1 Fig.

Discussion
This meta-analysis included 18 studies and 56 models aimed to investigate the performance of
deep learning in detecting AMD. The results of the present study indicate a high accuracy in
detecting AMD through CNN, but with high heterogeneity. The sources of heterogeneity were
the types of AMD and layers of the network according to the meta-regression.
DL has been widely adopted in image recognition, speech recognition, and natural language
processing, but is only beginning to impact healthcare, especially in ophthalmology [6]. DL is
a subset of machine learning which has become possible with increasing computing power.
Compared to traditional machine learning algorithms and shallow networks, current DL algo-
rithms are characterized by large amounts of processable data, high computational power, and
large network size [58, 59].
Fluorescein angiography, optical coherence tomography (OCT), optical coherence tomog-
raphy angiography (OCTA), FP, fundus autofluorescence, and indocyanine green angiography
are useful diagnostic tests in clinical practice to detect AMD [1], of which OCT and FP are the
most commonly used. Plenty of public ophthalmic datasets are based on the above two types
of images, which have facilitated the rapid development of artificial intelligence in ophthalmol-
ogy, and will make telemedicine more convenient in the future. This meta-analysis reveals that
DL detection through OCT, FP, and OPTOS ultra-widefield retinal images has a high accuracy
in AMD diagnosing.

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 12 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Fig 5. The SROC (a) the pooled SROC; (b) the SROC of types of AMD; (c) the SROC of types of images; (d) the SROC of variants of CNNs; (e) the SROC of
Networks.
https://doi.org/10.1371/journal.pone.0284060.g005

18 included studies were summarized in Table 2. All studies adopted CNN to conduct DL
models. The non-saturating ReLU activation function was introduced in AlexNet to increase
the training speed and the dropout method was used to minimize overfitting in the fully con-
nected layers [60]. VGG has a deeper architecture, but cannot overcome the limitation of the
vanishing gradient problem [61]. In the ResNet architecture, identity mapping is introduced
to solve the vanishing gradient problem. ResNet can therefore be used to train deeper models
[62]. DenseNets can alleviate the vanishing-gradient problem, strengthen feature propagation,
encourage feature reuse, and substantially reduce the number of parameters [63]. Other stud-
ies mostly used a self-created CNN architectures with 7–20 layers. Duke, Mendeley, and Noor
are the most used OCT databases. Most FP image datasets were built clinically. 10 studies [40,
41, 43, 46, 47, 52–55, 57] included more than 1000 images in their research. 4 studies [40, 43,
47, 57] included more than 100 thousand images. Only Matsuba et al [50] used OPTOS as the
dataset, which is unique and pioneering among the 18 studies. Dong et al [43] and Gour et al
[44] included 11 and 8 classes respectively, while other studies only contained 2–4 classes.
In this research, the type of AMD and the layers of the network were found to be the two
essential factors that impact the accuracy of the diagnosis. However, the layers of the network

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 13 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Table 3. Subgroup analyses and meta regression results.

Number of Sen1 Spe2 PLR NLR DOR AUC Meta
models regression
Pooled I2 Pooled I2 Pooled I2 Pooled I2 Pooled (95%CI) I2 P RDOR
(95%CI) (95%CI) (95%CI) (95%CI)
Layers 0.4878 0.74
�10 5 1.00 97.80% 0.99 60.20% 104.30 65.70% 0 (0.00– 98.60% 35609.58 93.80% 0.9992
(1.00– (0.99– (70.94– 0.03) (4947.73–
1.00) 0.99) 153.36) 256287.76)
10–20 28 0.91 99.70% 0.97 99.80% 7.40 (5.04– 99.20% 0.10 99.60% 52.47 (33.55– 98.50% 0.9861
(0.92– (0.97– 10.86) (0.07– 82.06)
0.91) 0.96) 0.16)
20–50 11 0.94 96.70% 0.95 97.80% 28.68 98.30% 0.04 96.40% 770.17 (163.74– 97.90% 0.9927
(0.93– (0.95– (10.64– (0.02– 3622.58)
0.95) 0.96) 77.28) 0.09)
50–100 7 0.93 85.40% 0.98 72.10% 36.07 50.80% 0.10 70.70% 366.71 (219.09– 59.70% 0.9935
(0.91– (0.97– (30.09– (0.06– 613.82)
0.94) 0.98) 43.24) 0.16)
>100 4 0.95 64.80% 0.98 93.70% 76.90 90.70% 0.04 57.10% 1967.84 84.40% 0.9914
(0.94– (0.97– (21.55– (0.03– (464.67–
0.97) 0.99) 274.39) 0.08) 8333.64)
Type of AMD 0.1882 36.03
Single 2 1.00 0% 0.99 89.0% 122.44 87.70% 0.00 0% 107120.15 0% NA3
(0.99– (0.98– (6.08– (0.00– (8471.87–
1.00) 1.00) 2463.89) 0.02) 1354450.49)
All 54 0.94 99.70% 0.97 99.70% 20.61 99.30% 0.06 99.70% 306.14 (187.38– 99.20% 0.9915
(0.94– (0.97– (14.59– (0.04– 500.18)
0.94) 0.97) 29.12) 0.09)
Architectures 0.0004 0.19
ResNet 13 0.94 96.10% 0.96 97.60% 34.66 98.20% 0.04 96.00% 951.18 (232.03– 97.70% 0.9941
(0.93– (0.95– (13.66– (0.02– 3899.24)
0.95) 0.96) 87.94) 0.08)
VGG 16 0.90 99.60% 0.91 98.50% 4.29 (3.25– 98.30% 0.19 99.50% 30.62 (21.91– 96.90% 0.8972
(0.90– (0.91– 5.67) (0.13– 42.79)
0.89) 0.90) 0.27)
Others 26 0.99 99.50% 0.99 99.40% 86.04 99.50% 0.01 99.70% 7354.18 99.40% 0.9987
(0.99– (0.99– (33.97– (0.00– (1231.62–
0.99) 0.99) 217.95) 0.08) 43913.09)
Types of 0.0002 0.12
Images
OCT 36 0.94 99.80% 0.97 99.70% 52.33 99.50% 0.03 99.70% 2209.74 99.40% 0.9982
(0.94– (0.97– (32.57– (0.02– (1113.02–
0.94) 0.97) 84.10) 0.04) 4387.10)
FP 11 0.83 96.90% 0.96 98.20% 9.26 (3.74– 99.10% 0.18 99.10% 51.02 (13.37– 97.70% 0.9592
(0.81– (0.96– 22.92) (0.06– 194.69)
0.84) 0.97) 0.53)
OCT & FP 8 0.84 45.70% 0.88 87.00% 4.38 (3.11– 91.30% 0.18 30.00% 28.80 (20.38– 70.70% 0.9176
(0.83– (0.87– 6.17) (0.15– 40.71)
0.85) 0.89) 0.20)
OPTOS 1 1.00 0% 0.97 0 34.95 0% 0 0 9371.15 0% NA
(0.97– (0.94– (20.45– (523.79–
1.00) 0.99) 59.73) 167661.33)

Sen: Sensitivity; Spe: Specificity; NA: not applicable.

https://doi.org/10.1371/journal.pone.0284060.t003

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 14 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

are not positively correlated with diagnostic accuracy. Even though DOR and AUC are higher
when the layers are less than 10, as layers of the network are more than 10, the diagnostic accu-
racy gradually grows as the number of layers increases. Cautiously, when the number of layers
becomes too deep, overfitting may occur. Overfitting is a serious issue when training DL mod-
els, which may cause the trained models cannot be generalized in other data or datasets [64].
Predictably, deeper and more accurate networks will be placed in service soon. Meanwhile, dif-
ferent types of AMD may make the computation more difficult, but the prevailing datasets
contain different types of AMD, which will make the trained models more generalized.
Although the meta-regression results did not show that the networks and types of images
connected to the diagnostic accuracy, they are still significant. The DOR of ResNet showed
superior than VGG, other variants cannot be assessed because they were only included in one
study. That might be because ResNet with more layers was developed after VGG. ResNet
belongs to deep residual networks with a higher amount of processable data [65]. ResNets can
be trained easily without increasing the training error percentage, and are helpful in tackling
the vanishing gradient problem using identity mapping [66]. Therefore, it is believed ResNet is
an ideal architecture among the present variants of CNN. However, the influence due to the
layers of the network impacts the results. This may be the reason the RDOR of networks in
meta-regression is very low. As for the types of images, OCT images showed superior in
detecting AMD. OCT images can reveal every layer of macular structures with more anatomi-
cal information than fundus images. Combined OCT images with fundus images had worse
sensitivity, specificity, DOR, and AUC. We think it is because two images have more informa-
tion. More information means more computation and the potential to be more accurate,
which may require considering the layers of the network and the architecture of CNNs. Addi-
tionally, although only one study [50] reported the OPTOS ultra-widefield retinal images as
self-dataset, the pooled sensitivity, specificity, and DOR were all highest in the four subgroups.
The detailed limitations for each study were summarized in Table 3. Generally, 8 studies
[42, 47, 50–54, 56] did not study other retinal diseases. 3 studies [45, 49, 55] only contained
one other diseases. 2 studies [48, 56] had small datasets with no more than 100 images. Bhatia
et al [41] ignored bad quality pictures that may cause a generalization issue. Celebi et al [42]
ignored bad quality pictures and patients who had other retinal diseases. Although Dong et al
[43] established a database with 11 classes, the number of retinitis pigmentosa images is small.
Gour et al [44] contained 8 types of diseases, but with a small dataset of 331 FP images. Kadry
et al [46] used 4 CNN variants for classification, but the definition of non-AMD is not clear.
Lee et al [47] included only images from patients who met the study criteria, and the neural
network was only trained on these images. Meanwhile, the model was trained using images
from a single academic center, and the external generalizability is unknown. Mathews et al
[49] used drusen macular degeneration for AMD diagnosis. Matsuba et al [50] used OPTOS
images, but it is difficult to acquire precise images using OPTOS when the transmission of
light into the eye is impaired by an intermediate translucent zone. At the same time, most
AMD patients accept treatment which may cause diagnostic errors. Motozawa et al [51]
excluded low quality images and patients who had other concomitant diseases. Zapata et al did
not report a clear number of OCT or FP images.
This meta-analysis and the included studies have several limitations. First, some variants of
CNN including CapsNet, AlexNet, and DenseNet only existed once, and some studies used
self-created CNN architectures. Therefore, the subgroup analysis of networks is not accurate.
Second, we tried to establish more subgroups or to find more possible covariates such as hard-
ware, network, and hyperparameters. However, these potential factors were not mentioned in
many studies. Third, we concentrated more on diagnostic accuracy, but as DL develops, AMD
diagnostics has become more diverse, more plentiful, and more useful in lesion segmentation

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 15 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

and efficacy prediction, which will be highly considered for further research. Forth, the Duke
and some self-built datasets have a small number of images for training.

Future challenges and direction

DL is still in the early stages of development in AMD diagnosis, but in the foreseeable future,
widespread use could play a significant role in fundus applications, screening, telemedicine,
reducing the workload of physicians, etc.
The purpose of DL algorithms for diagnosing AMD is to achieve an automated diagnosis of
many kinds of fundus diseases. However, no matter in public databases or self-built databases,
only several diseases were chosen for classification which is difficult for widespread use clini-
cally. Establishing a database which covers heterogeneous and large image sets is still a serious
challenge. Meanwhile, the DL algorithms concentrate more on images only, but the images are
not the only data obtained clinically. Ideally, multimodal data containing clinical data, FP, and
OCT, etc. may increase the diagnostic accuracy. At the same time, traditional fundus datasets
mostly consist of FP and OCT images. However, with diagnostic tests developing, more new
methods and technologies such as OPTOS ultra-widefield retinal images, OCTA, FFA, ICGA,
etc. will be added as public or self-built datasets in future AMD detection. Finally, as the equip-
ment evolves, the image quality of FP, OCT, OCTA etc. improves. More high definition images
will increase the diagnostic accuracy.

Conclusions
CNNs are mostly adopted deep learning algorithms in AMD detection. All included DL algo-
rithms adopted CNNs. CNNs, especially ResNets, are effective in detecting AMD with high
diagnostic accuracy. The types of AMD and the layers of the network are the two essential fac-
tors that impact the model training process. Proper layers of the network will make the model
more reliable. More datasets established by new diagnostic methods such as ultra-widefield
retinal images, FFA, and ICGA will be used to train DL models in the future, which will be
helpful in fundus application screening, long-range medical treatment, and reducing the work-
load of physicians.

Supporting information
S1 Fig. The original forest plots.
(PDF)
S1 Table. PRISMA checklist.
(DOC)
S2 Table. PRISMA DTA for abstracts checklist table.
(DOC)
S1 File. Search strategy.
(DOCX)

Author Contributions
Conceptualization: Xiangjie Leng, Xuejing Lu.
Data curation: Xiangjie Leng, Ruijie Shi, Yanxia Wu.
Formal analysis: Xiangjie Leng, Ruijie Shi.

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 16 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

Funding acquisition: Xuejing Lu.

Investigation: Xiangjie Leng, Ruijie Shi, Shiyin Zhu, Xingcan Cai.
Methodology: Xiangjie Leng, Ruijie Shi, Yanxia Wu, Shiyin Zhu, Xingcan Cai.
Software: Xiangjie Leng, Ruijie Shi.
Supervision: Xuejing Lu.
Validation: Yanxia Wu, Shiyin Zhu, Xingcan Cai.
Writing – original draft: Xiangjie Leng, Ruijie Shi, Yanxia Wu.
Writing – review & editing: Xiangjie Leng, Ruijie Shi, Yanxia Wu, Xuejing Lu, Ruobing Liu.

References
1. Flaxel CJ, Adelman RA, Bailey ST, Fawzi A, Lim JI, Vemulakonda GA, et al. Age-Related Macular
Degeneration Preferred Practice Pattern(R). Ophthalmology. 2020; 127(1):P1–P65. Epub 2019/11/24.
https://doi.org/10.1016/j.ophtha.2019.09.024 PMID: 31757502.
2. Congdon N, O’Colmain B, Klaver CC, Klein R, Muñoz B, Friedman DS, et al. Causes and prevalence of
visual impairment among adults in the United States. Archives of ophthalmology (Chicago, Ill: 1960).
2004; 122(4):477–85. Epub 2004/04/14. https://doi.org/10.1001/archopht.122.4.477 PMID: 15078664.
3. Wong WL, Su X, Li X, Cheung CMG, Klein R, Cheng C-Y, et al. Global prevalence of age-related macu-
lar degeneration and disease burden projection for 2020 and 2040: a systematic review and meta-anal-
ysis. The Lancet Global Health. 2014; 2(2):e106–e16. https://doi.org/10.1016/S2214-109X(13)70145-1
PMID: 25104651
4. Mitchell P, Liew G, Gopinath B, Wong TY. Age-related macular degeneration. Lancet (London,
England). 2018; 392(10153):1147–59. Epub 2018/10/12. https://doi.org/10.1016/S0140-6736(18)
31550-2 PMID: 30303083.
5. Tsui I, Jain A, Shah S, Schwartz SD, McCannel TA, editors. Ultra widefield imaging of peripheral exuda-
tive hemorrhagic chorioretinopathy. Seminars in ophthalmology; 2009: Taylor & Francis.
6. Nagiel A, Lalane RA, Sadda SR, Schwartz SDJR. Ultra-widefield fundus imaging: a review of clinical
applications and future trends. 2016; 36(4):660–78. https://doi.org/10.1097/IAE.0000000000000937
PMID: 27014860
7. Schmidt-Erfurth U, Sadeghipour A, Gerendas BS, Waldstein SM, Bogunovic H. Artificial intelligence in
retina. Prog Retin Eye Res. 2018; 67:1–29. Epub 2018/08/05. https://doi.org/10.1016/j.preteyeres.
2018.07.004 PMID: 30076935.
8. Ting DSW, Pasquale LR, Peng L, Campbell JP, Lee AY, Raman R, et al. Artificial intelligence and deep
learning in ophthalmology. Br J Ophthalmol. 2019; 103(2):167–75. Epub 2018/10/27. https://doi.org/10.
1136/bjophthalmol-2018-313173 PMID: 30361278; PubMed Central PMCID: PMC6362807.
9. Kermany DS, Goldbaum M, Cai W, Valentim CCS, Liang H, Baxter SL, et al. Identifying Medical Diagno-
ses and Treatable Diseases by Image-Based Deep Learning. Cell. 2018; 172(5):1122–31 e9. Epub
2018/02/24. https://doi.org/10.1016/j.cell.2018.02.010 PMID: 29474911.
10. Chan S, Reddy V, Myers B, Thibodeaux Q, Brownstone N, Liao W. Machine Learning in Dermatology:
Current Applications, Opportunities, and Limitations. Dermatol Ther (Heidelb). 2020; 10(3):365–86.
Epub 2020/04/08. https://doi.org/10.1007/s13555-020-00372-0 PMID: 32253623; PubMed Central
PMCID: PMC7211783.
11. Saba L, Biswas M, Kuppili V, Cuadrado Godia E, Suri HS, Edla DR, et al. The present and future of
deep learning in radiology. Eur J Radiol. 2019; 114:14–24. Epub 2019/04/22. https://doi.org/10.1016/j.
ejrad.2019.02.038 PMID: 31005165.
12. McBee MP, Awan OA, Colucci AT, Ghobadi CW, Kadom N, Kansagra AP, et al. Deep Learning in Radi-
ology. Acad Radiol. 2018; 25(11):1472–80. Epub 2018/04/03. https://doi.org/10.1016/j.acra.2018.02.
018 PMID: 29606338.
13. Jiang Y, Yang M, Wang S, Li X, Sun Y. Emerging role of deep learning-based artificial intelligence in
tumor pathology. Cancer Commun (Lond). 2020; 40(4):154–66. Epub 2020/04/12. https://doi.org/10.
1002/cac2.12012 PMID: 32277744; PubMed Central PMCID: PMC7170661.
14. Wang S, Yang DM, Rong R, Zhan X, Xiao G. Pathology Image Analysis Using Segmentation Deep
Learning Algorithms. Am J Pathol. 2019; 189(9):1686–98. Epub 2019/06/15. https://doi.org/10.1016/j.
ajpath.2019.05.007 PMID: 31199919; PubMed Central PMCID: PMC6723214.

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 17 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

15. Ting DSW, Cheung CY, Lim G, Tan GSW, Quang ND, Gan A, 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. JAMA. 2017; 318(22):2211–23. Epub 2017/12/14. https://doi.
org/10.1001/jama.2017.18152 PMID: 29234807; PubMed Central PMCID: PMC5820739.
16. Abramoff MD, Lou Y, Erginay A, Clarida W, Amelon R, Folk JC, et al. Improved Automated Detection of
Diabetic Retinopathy on a Publicly Available Dataset Through Integration of Deep Learning. Invest
Ophthalmol Vis Sci. 2016; 57(13):5200–6. Epub 2016/10/05. https://doi.org/10.1167/iovs.16-19964
PMID: 27701631.
17. Sahlsten J, Jaskari J, Kivinen J, Turunen L, Jaanio E, Hietala K, et al. Deep Learning Fundus Image
Analysis for Diabetic Retinopathy and Macular Edema Grading. Sci Rep. 2019; 9(1):10750. Epub 2019/
07/26. https://doi.org/10.1038/s41598-019-47181-w PMID: 31341220; PubMed Central PMCID:
PMC6656880.
18. Grassmann F, Mengelkamp J, Brandl C, Harsch S, Zimmermann ME, Linkohr B, et al. A Deep Learning
Algorithm for Prediction of Age-Related Eye Disease Study Severity Scale for Age-Related Macular
Degeneration from Color Fundus Photography. Ophthalmology. 2018; 125(9):1410–20. Epub 2018/04/
15. https://doi.org/10.1016/j.ophtha.2018.02.037 PMID: 29653860.
19. Rim TH, Lee AY, Ting DS, Teo K, Betzler BK, Teo ZL, et al. Detection of features associated with neo-
vascular age-related macular degeneration in ethnically distinct data sets by an optical coherence
tomography: trained deep learning algorithm. Br J Ophthalmol. 2021; 105(8):1133–9. Epub 2020/09/11.
https://doi.org/10.1136/bjophthalmol-2020-316984 PMID: 32907811; PubMed Central PMCID:
PMC8185637.
20. Waldstein SM, Vogl WD, Bogunovic H, Sadeghipour A, Riedl S, Schmidt-Erfurth U. Characterization of
Drusen and Hyperreflective Foci as Biomarkers for Disease Progression in Age-Related Macular
Degeneration Using Artificial Intelligence in Optical Coherence Tomography. JAMA Ophthalmol. 2020;
138(7):740–7. Epub 2020/05/08. https://doi.org/10.1001/jamaophthalmol.2020.1376 PMID: 32379287;
PubMed Central PMCID: PMC7206537.
21. Bhuiyan A, Govindaiah A, Smith RT. An Artificial-Intelligence- and Telemedicine-Based Screening Tool
to Identify Glaucoma Suspects from Color Fundus Imaging. J Ophthalmol. 2021; 2021:6694784. Epub
2021/06/18. https://doi.org/10.1155/2021/6694784 PMID: 34136281; PubMed Central PMCID:
PMC8179760 conflicts of interest.
22. Varadarajan AV, Poplin R, Blumer K, Angermueller C, Ledsam J, Chopra R, et al. Deep Learning for
Predicting Refractive Error From Retinal Fundus Images. Invest Ophthalmol Vis Sci. 2018; 59(7):2861–
8. Epub 2018/07/20. https://doi.org/10.1167/iovs.18-23887 PMID: 30025129.
23. Brown JM, Campbell JP, Beers A, Chang K, Ostmo S, Chan RVP, et al. Automated Diagnosis of Plus
Disease in Retinopathy of Prematurity Using Deep Convolutional Neural Networks. JAMA Ophthalmol.
2018; 136(7):803–10. Epub 2018/05/26. https://doi.org/10.1001/jamaophthalmol.2018.1934 PMID:
29801159; PubMed Central PMCID: PMC6136045.
24. Campbell JP, Kim SJ, Brown JM, Ostmo S, Chan RVP, Kalpathy-Cramer J, et al. Evaluation of a Deep
Learning-Derived Quantitative Retinopathy of Prematurity Severity Scale. Ophthalmology. 2021; 128
(7):1070–6. Epub 2020/10/31. https://doi.org/10.1016/j.ophtha.2020.10.025 PMID: 33121959; PubMed
Central PMCID: PMC8076329.
25. Gensure RH, Chiang MF, Campbell JP. Artificial intelligence for retinopathy of prematurity. Curr Opin
Ophthalmol. 2020; 31(5):312–7. Epub 2020/07/23. https://doi.org/10.1097/ICU.0000000000000680
PMID: 32694266; PubMed Central PMCID: PMC7891849.
26. Khan N, Adam R, Huang P, Maldjian T, Duong TQ. Deep Learning Prediction of Pathologic Complete
Response in Breast Cancer Using MRI and Other Clinical Data: A Systematic Review. Tomography.
2022; 8(6):2784–95. Epub 20221121. https://doi.org/10.3390/tomography8060232 PMID: 36412691.
27. Ting DSW, Peng L, Varadarajan AV, Keane PA, Burlina PM, Chiang MF, et al. Deep learning in ophthal-
mology: The technical and clinical considerations. Prog Retin Eye Res. 2019; 72:100759. Epub 2019/
05/03. https://doi.org/10.1016/j.preteyeres.2019.04.003 PMID: 31048019.
28. Peking university international competition on ocular disease intelligent recognition (ODIR-2019) 2020
[cited 2022 December 28, 2022]. Available from: https://odir2019.grand-challenge.org/dataset/.
29. grand-challenge.org. iChallenge-AMD 2021 [December 28, 2022]. Available from: https://refuge.grand-
challenge.org/iChallenge-AMD/.
30. Srinivasan PP, Kim LA, Mettu PS, Cousins SW, Comer GM, Izatt JA, et al. Fully automated detection of
diabetic macular edema and dry age-related macular degeneration from optical coherence tomography
images. Biomedical optics express. 2014; 5(10):3568–77. https://doi.org/10.1364/BOE.5.003568
PMID: 25360373
31. Rasti R, Rabbani H, Mehridehnavi A, Hajizadeh F. Macular OCT classification using a multi-scale con-
volutional neural network ensemble. IEEE transactions on medical imaging. 2017; 37(4):1024–34.

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 18 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

32. Mooney P. Retinal OCT Images (optical coherence tomography) 2017 [December 28, 2022]. Available
from: https://www.kaggle.com/datasets/paultimothymooney/kermany2018.
33. Kermany D, Zhang K, Goldbaum M. Labeled optical coherence tomography (oct) and chest x-ray
images for classification. Mendeley data. 2018; 2(2).
34. Gholami P, Roy P, Parthasarathy MK, Lakshminarayanan V. OCTID: Optical coherence tomography
image database. Computers & Electrical Engineering. 2020; 81:106532.
35. McInnes MDF, Moher D, Thombs BD, McGrath TA, Bossuyt PM, and the P-DTAG, et al. Preferred
Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies: The
PRISMA-DTA Statement. JAMA. 2018; 319(4):388–96. Epub 2018/01/25. https://doi.org/10.1001/
jama.2017.19163 PMID: 29362800.
36. Leeflang MM, Deeks JJ, Takwoingi Y, Macaskill PJSr. Cochrane diagnostic test accuracy reviews.
2013; 2(1):1–6. https://doi.org/10.1186/2046-4053-2-82 PMID: 24099098
37. Liang S, Chang LJBim. Serum matrix metalloproteinase-9 level as a biomarker for colorectal cancer: a
diagnostic meta-analysis. 2018; 12(4):393–402. https://doi.org/10.2217/bmm-2017-0206 PMID:
29575908
38. Higgins JP, Thompson SG, Deeks JJ, Altman DGJB. Measuring inconsistency in meta-analyses. 2003;
327(7414):557–60. https://doi.org/10.1136/bmj.327.7414.557 PMID: 12958120
39. Higgins JP, Thompson SGJSim. Quantifying heterogeneity in a meta-analysis. 2002; 21(11):1539–58.
https://doi.org/10.1002/sim.1186 PMID: 12111919
40. Alqudah AM. AOCT-NET: a convolutional network automated classification of multiclass retinal dis-
eases using spectral-domain optical coherence tomography images. Med Biol Eng Comput. 2020; 58
(1):41–53. Epub 2019/11/16. https://doi.org/10.1007/s11517-019-02066-y PMID: 31728935.
41. Bhatia KK, Graham MS, Terry L, Wood A, Tranos P, Trikha S, et al. DISEASE CLASSIFICATION OF
MACULAR OPTICAL COHERENCE TOMOGRAPHY SCANS USING DEEP LEARNING SOFT-
WARE: Validation on Independent, Multicenter Data. Retina (Philadelphia, Pa). 2020; 40(8):1549–57.
Epub 2019/10/05. https://doi.org/10.1097/IAE.0000000000002640 PMID: 31584557.
42. Celebi ARC, Bulut E, Sezer A. Artificial intelligence based detection of age-related macular degenera-
tion using optical coherence tomography with unique image preprocessing. Eur J Ophthalmol.
2022:11206721221096294. Epub 2022/04/27. https://doi.org/10.1177/11206721221096294 PMID:
35469472.
43. Dong L, He W, Zhang R, Ge Z, Wang YX, Zhou J, et al. Artificial Intelligence for Screening of Multiple
Retinal and Optic Nerve Diseases. JAMA Netw Open. 2022; 5(5):e229960. Epub 2022/05/04. https://
doi.org/10.1001/jamanetworkopen.2022.9960 PMID: 35503220; PubMed Central PMCID:
PMC9066285.
44. Gour N, Khanna P. Multi-class multi-label ophthalmological disease detection using transfer learning
based convolutional neural network. Biomedical Signal Processing and Control. 2021; 66. https://doi.
org/10.1016/j.bspc.2020.102329
45. He T, Zhou Q, Zou Y. Automatic Detection of Age-Related Macular Degeneration Based on Deep
Learning and Local Outlier Factor Algorithm. Diagnostics (Basel). 2022; 12(2). Epub 2022/02/26.
https://doi.org/10.3390/diagnostics12020532 PMID: 35204621; PubMed Central PMCID:
PMC8871377.
46. Kadry S, Rajinikanth V, Crespo RG, Verdu E. Automated detection of age-related macular degeneration
using a pre-trained deep-learning scheme. JOURNAL OF SUPERCOMPUTING. 2022; 78(5):7321–40.
https://doi.org/10.1007/s11227-021-04181-w WOS:000716301900001.
47. Lee CS, Baughman DM, Lee AY. Deep learning is effective for the classification of OCT images of nor-
mal versus Age-related Macular Degeneration. Ophthalmol Retina. 2017; 1(4):322–7. Epub 2017/07/
01. https://doi.org/10.1016/j.oret.2016.12.009 PMID: 30693348; PubMed Central PMCID:
PMC6347658.
48. Ma D, Kumar M, Khetan V, Sen P, Bhende M, Chen S, et al. Clinical explainable differential diagnosis of
polypoidal choroidal vasculopathy and age-related macular degeneration using deep learning. Comput
Biol Med. 2022; 143:105319. Epub 2022/02/28. https://doi.org/10.1016/j.compbiomed.2022.105319
PMID: 35220077.
49. Mathews MR, Anzar STM. A lightweight deep learning model for retinal optical coherence tomography
image classification. INTERNATIONAL JOURNAL OF IMAGING SYSTEMS AND TECHNOLOGY.
https://doi.org/10.1002/ima.22808 WOS:000856143200001.
50. Matsuba S, Tabuchi H, Ohsugi H, Enno H, Ishitobi N, Masumoto H, et al. Accuracy of ultra-wide-field
fundus ophthalmoscopy-assisted deep learning, a machine-learning technology, for detecting age-
related macular degeneration. Int Ophthalmol. 2019; 39(6):1269–75. Epub 2018/05/11. https://doi.org/
10.1007/s10792-018-0940-0 PMID: 29744763.

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 19 / 20

PLOS ONE Deep learning for detection of age-related macular degeneration

51. Motozawa N, An G, Takagi S, Kitahata S, Mandai M, Hirami Y, et al. Optical Coherence Tomography-
Based Deep-Learning Models for Classifying Normal and Age-Related Macular Degeneration and Exu-
dative and Non-Exudative Age-Related Macular Degeneration Changes. Ophthalmol Ther. 2019; 8
(4):527–39. Epub 2019/08/14. https://doi.org/10.1007/s40123-019-00207-y PMID: 31407214; PubMed
Central PMCID: PMC6858411.
52. Takhchidi K, Gliznitsa PV, Svetozarskiy SN, Bursov AI, Shusterzon KA. Labelling of data on fundus
color pictures used to train a deep learning model enhances its macular pathology recognition capabili-
ties. Bulletin of Russian State Medical University. 2021;( 2021(4)). https://doi.org/10.24075/brsmu.
2021.040
53. Tan JH, Bhandary SV, Sivaprasad S, Hagiwara Y, Bagchi A, Raghavendra U, et al. Age-related Macular
Degeneration detection using deep convolutional neural network. Future Generation Computer Sys-
tems. 2018; 87:127–35. https://doi.org/10.1016/j.future.2018.05.001.
54. Thomas A, Harikrishnan PM, Gopi VP, Palanisamy P. An Automated Method to Detect Age-Related
Macular Degeneration from Optical Coherence Tomographic Images. Biomedical Engineering: Applica-
tions, Basis and Communications. 2021; 33(05). https://doi.org/10.4015/s1016237221500368
55. Wang DP, Wang LJ. On OCT Image Classification via Deep Learning. IEEE PHOTONICS JOURNAL.
2019; 11(5). https://doi.org/10.1109/JPHOT.2019.2934484 WOS:000484194500001.
56. Yoo TK, Choi JY, Seo JG, Ramasubramanian B, Selvaperumal S, Kim DW. The possibility of the combi-
nation of OCT and fundus images for improving the diagnostic accuracy of deep learning for age-related
macular degeneration: a preliminary experiment. Med Biol Eng Comput. 2019; 57(3):677–87. Epub
2018/10/24. https://doi.org/10.1007/s11517-018-1915-z PMID: 30349958.
57. Zapata MA, Royo-Fibla D, Font O, Vela JI, Marcantonio I, Moya-Sanchez EU, et al. Artificial Intelligence
to Identify Retinal Fundus Images, Quality Validation, Laterality Evaluation, Macular Degeneration, and
Suspected Glaucoma. Clin Ophthalmol. 2020; 14:419–29. Epub 2020/02/28. https://doi.org/10.2147/
OPTH.S235751 PMID: 32103888; PubMed Central PMCID: PMC7025650.
58. Survarachakan S, Prasad PJR, Naseem R, Perez de Frutos J, Kumar RP, Lango T, et al. Deep learning
for image-based liver analysis—A comprehensive review focusing on malignant lesions. Artif Intell Med.
2022; 130:102331. Epub 2022/07/10. https://doi.org/10.1016/j.artmed.2022.102331 PMID: 35809970.
59. Romond K, Alam M, Kravets S, Sisternes L, Leng T, Lim JI, et al. Imaging and artificial intelligence for
progression of age-related macular degeneration. Exp Biol Med (Maywood). 2021; 246(20):2159–69.
Epub 2021/08/19. https://doi.org/10.1177/15353702211031547 PMID: 34404252; PubMed Central
PMCID: PMC8718252.
60. Krizhevsky A, Sutskever I, Hinton GE. Imagenet classification with deep convolutional neural networks.
Communications of the ACM. 2017; 60(6):84–90.
61. Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. arXiv
preprint arXiv:14091556. 2014.
62. He K, Zhang X, Ren S, Sun J, editors. Deep residual learning for image recognition. Proceedings of the
IEEE conference on computer vision and pattern recognition; 2016.
63. Zhu Y, Newsam S, editors. Densenet for dense flow. 2017 IEEE international conference on image pro-
cessing (ICIP); 2017: IEEE.
64. Xie Z, He F, Fu S, Sato I, Tao D, Sugiyama M. Artificial Neural Variability for Deep Learning: On Overfit-
ting, Noise Memorization, and Catastrophic Forgetting. Neural computation. 2021; 33(8):2163–92.
Epub 2021/07/27. https://doi.org/10.1162/neco_a_01403 PMID: 34310675.
65. Alzubaidi L, Zhang J, Humaidi AJ, Al-Dujaili A, Duan Y, Al-Shamma O, et al. Review of deep learning:
Concepts, CNN architectures, challenges, applications, future directions. 2021; 8(1):1–74. https://doi.
org/10.1186/s40537-021-00444-8 PMID: 33816053
66. Chen Z, Xie Z, Zhang W, Xu X, editors. ResNet and Model Fusion for Automatic Spoofing Detection.
Interspeech; 2017.

PLOS ONE | https://doi.org/10.1371/journal.pone.0284060 April 6, 2023 20 / 20