Synthetic artificial intelligence using generative adversarial network for retinal imaging in detection of age-related macular degeneration.

Introduction: Age-related macular degeneration (AMD) is one of the leading causes of vision impairment globally and early detection is crucial to prevent vision loss. However, the screening of AMD is resource dependent and demands experienced healthcare providers. Recently, deep learning (DL) systems have shown the potential for effective detection of various eye diseases from retinal fundus images, but the development of such robust systems requires a large amount of datasets, which could be limited by prevalence of the disease and privacy of patient. As in the case of AMD, the advanced phenotype is often scarce for conducting DL analysis, which may be tackled via generating synthetic images using Generative Adversarial Networks (GANs). This study aims to develop GAN-synthesized fundus photos with AMD lesions, and to assess the realness of these images with an objective scale. Methods: To build our GAN models, a total of 125,012 fundus photos were used from a real-world non-AMD phenotypical dataset. StyleGAN2 and human-in-the-loop (HITL) method were then applied to synthesize fundus images with AMD features. To objectively assess the quality of the synthesized images, we proposed a novel realness scale based on the frequency of the broken vessels observed in the fundus photos. Four residents conducted two rounds of gradings on 300 images to distinguish real from synthetic images, based on their subjective impression and the objective scale respectively. Results and discussion: The introduction of HITL training increased the percentage of synthetic images with AMD lesions, despite the limited number of AMD images in the initial training dataset. Qualitatively, the synthesized images have been proven to be robust in that our residents had limited ability to distinguish real from synthetic ones, as evidenced by an overall accuracy of 0.66 (95% CI: 0.61-0.66) and Cohen's kappa of 0.320. For the non-referable AMD classes (no or early AMD), the accuracy was only 0.51. With the objective scale, the overall accuracy improved to 0.72. In conclusion, GAN models built with HITL training are capable of producing realistic-looking fundus images that could fool human experts, while our objective realness scale based on broken vessels can help identifying the synthetic fundus photos.

Understanding patient preferences in anti-VEGF treatment options for age-related macular degeneration.

PURPOSE: (1) To investigate the relative importance of convenience (consultation frequency and injection frequency) against treatment outcomes (visual and anatomical outcomes) and out-of-pocket medical costs via a discrete choice experiment (DCE), and (2) to investigate how patient characteristics affect patient treatment preferences. METHODS: Eligibility criteria were: (1) receiving a neovascular age-related macular degeneration (nAMD) diagnosis; (2) receiving anti-VEGF treatment; (3) being ≥21 years old, and (4) being able to speak and understand English/Mandarin. Patients were presented with eight choice tasks and asked to choose between their current treatment and two hypothetical treatments that varied by six attributes: number of clinic visits in a year, number of injections in a year, vision quality, control of swelling in retina, drug labelling and out-of-pocket cost. RESULTS: This analysis involved 180 patients. Based on latent class logistic regressions, vision quality was the most important attribute (34%) followed by cost (24%). The frequency of total clinic visits (15%) was the third most-important attribute, closely followed by labelling (12%) and control of retina swelling (11%). Injection frequency was the least important attribute (4%). CONCLUSIONS: Vision quality was the most important attribute followed by the out-of-pocket costs. Given the same outcomes, patients preferred treatment regimens which require fewer total clinic visits. In comparison, injection frequency alone did not influence patient preferences. With increasing treatment options for nAMD, understanding patients' preferences can help clinicians in selecting agents and treatment regimen most preferred for each patient, which may lead to improved long-term adherence and outcomes.

Impact of incident age-related macular degeneration and associated vision loss on vision-related quality of life.

BACKGROUND: We examined the associations between the 6-year incidence of age-related macular degeneration (AMD) and vision-related quality of life (VRQoL), and the contribution of presenting visual acuity (VA), in an Asian population. METHODS: Fundus images from the Singapore Chinese Eye Study, a population-based cohort study (baseline: 2009-2011; follow-up: 2015-2017), were graded using a modified Wisconsin age-related maculopathy grading system. Incident AMD was defined as no baseline AMD in both eyes and early/late AMD in the worse eye at follow-up. Presenting VA was assessed using the logarithm of the minimum angle of resolution chart at 4 m under standard lighting conditions with habitual correction. Multiple linear regression models determined the associations between AMD incidence with changes in the Rasch-transformed scores of the Reading, Mobility and Emotional VRQoL domains of the 32-item Impact of Visual Impairment (IVI-32) questionnaire, adjusted for traditional confounders. The contribution of presenting VA to changes in VRQoL was also estimated. RESULTS: Of the 2251 participants without AMD at baseline (mean age (SD): 57.7 (9) years, 51.4% women), 101 (4.5%) and 11 (0.5%) developed incident early and late AMD at follow-up, respectively. Incident late AMD was associated with significant 30.3%, 32.5% and 30.9% decrements in Reading, Mobility and Emotional IVI scores, respectively. The contribution of presenting VA ranged between 1.62% and 4.35% of the observed decrements. No significant associations were noted with incident early AMD. CONCLUSION: Incident late AMD had a substantial impact on all aspects of VRQoL, with presenting VA contributing only minimally to this longitudinal relationship.

OPTICAL COHERENCE TOMOGRAPHY FEATURES OF POLYPOIDAL LESION CLOSURE IN POLYPOIDAL CHOROIDAL VASCULOPATHY TREATED WITH AFLIBERCEPT.

PURPOSE: To evaluate whether optical coherence tomography (OCT) can determine polypoidal lesion (PL) perfusion in polypoidal choroidal vasculopathy eyes after 12 months of aflibercept monotherapy. Polypoidal lesion perfusion status, assessed by indocyanine green angiography, is an important anatomical outcome in polypoidal choroidal vasculopathy management. METHODS: Post hoc data from a prospective randomized, open-label, study in eyes with polypoidal choroidal vasculopathy undergoing monotherapy with aflibercept evaluated PL perfusion status based on indocyanine green angiography (gold standard) and OCT features from baseline to 12 months. RESULTS: Individual PLs (110 in total) from 48 eyes (48 patients) showed at 12 months; 57/110 PLs (51.8%) were closed on indocyanine green angiography. At 12 months, eyes with closed PLs were more likely to have the following OCT features: 1) no subretinal fluid (67.1% vs. 32.9%), 2) smaller pigment epithelial detachment height (67.2 [±43.8] vs. 189.2 [±104.9] µm), 3) densely hyperreflective pigment epithelial detachment contents (84.0% vs. 16.0%), 4) an absence of a hyperreflective ring(64.0% vs. 36.0%), and a 5) indistinct overlying retinal pigment epithelial (71.4% vs. 28.6%) (all P < 0.05). The three highest performing OCT features that differentiated perfused from closed PLs were (1), (3), and (4) (area under the receiver operating characteristic curve 0.85, 0.73, and 0.70, respectively). A combination of these three features achieved an area under the receiver operating characteristic curve of 0.90. CONCLUSION: Polypoidal lesion closure, an important anatomical treatment outcome in polypoidal choroidal vasculopathy typically defined by indocyanine green angiography, can be accurately detected by specific OCT features.

Retinal photograph-based deep learning algorithms for myopia and a blockchain platform to facilitate artificial intelligence medical research: a retrospective multicohort study.

BACKGROUND: By 2050, almost 5 billion people globally are projected to have myopia, of whom 20% are likely to have high myopia with clinically significant risk of sight-threatening complications such as myopic macular degeneration. These are diagnoses that typically require specialist assessment or measurement with multiple unconnected pieces of equipment. Artificial intelligence (AI) approaches might be effective for risk stratification and to identify individuals at highest risk of visual loss. However, unresolved challenges for AI medical studies remain, including paucity of transparency, auditability, and traceability. METHODS: In this retrospective multicohort study, we developed and tested retinal photograph-based deep learning algorithms for detection of myopic macular degeneration and high myopia, using a total of 226 686 retinal images. First we trained and internally validated the algorithms on datasets from Singapore, and then externally tested them on datasets from China, Taiwan, India, Russia, and the UK. We also compared the performance of the deep learning algorithms against six human experts in the grading of a randomly selected dataset of 400 images from the external datasets. As proof of concept, we used a blockchain-based AI platform to demonstrate the real-world application of secure data transfer, model transfer, and model testing across three sites in Singapore and China. FINDINGS: The deep learning algorithms showed robust diagnostic performance with areas under the receiver operating characteristic curves [AUC] of 0·969 (95% CI 0·959-0·977) or higher for myopic macular degeneration and 0·913 (0·906-0·920) or higher for high myopia across the external testing datasets with available data. In the randomly selected dataset, the deep learning algorithms outperformed all six expert graders in detection of each condition (AUC of 0·978 [0·957-0·994] for myopic macular degeneration and 0·973 [0·941-0·995] for high myopia). We also successfully used blockchain technology for data transfer, model transfer, and model testing between sites and across two countries. INTERPRETATION: Deep learning algorithms can be effective tools for risk stratification and screening of myopic macular degeneration and high myopia among the large global population with myopia. The blockchain platform developed here could potentially serve as a trusted platform for performance testing of future AI models in medicine. FUNDING: None.

Patterns and Determinants of Choroidal Thickness in a Multiethnic Asian Population: The Singapore Epidemiology of Eye Diseases Study.

PURPOSE: To describe the distribution and determinants of choroidal thickness (CT) in participants in a population study based on spectral-domain (SD)-OCT measurements. DESIGN: Population-based, cross-sectional study. PARTICIPANTS: Ethnic Chinese, Indian, and Malay adults aged more than 50 years without any retinal diseases (e.g., diabetic retinopathy, macular edema, age-related macular degeneration, central serous chorioretinopathy) that might affect the CT were recruited from the Singapore Epidemiology of Eye Diseases Study. METHODS: Choroidal imaging was performed by SD-OCT (Spectralis, Heidelberg Engineering, Heidelberg, Germany) in enhanced depth imaging (EDI) mode. Subfoveal choroidal thickness (SFCT) was measured on the foveal line scan by 2 retinal experts independently (YS and KT), and the average was used in the analyses. In Chinese and Indian cohorts in whom macular raster scans were captured, the manufacturer-supplied research software (Heyex SP-X version 6.4.8.116; Heidelberg Engineering) was used to obtain automated segmentation yielding mean choroidal thickness in each of the 9 ETDRS grid sectors. MAIN OUTCOME MEASURES: Subfoveal choroidal thickness and regional CT in the 9 ETDRS grid sectors. RESULTS: For the SFCT analysis, 2794 eyes of 1619 participants (Chinese, Indian, and Malay ) were included. The mean age was 60.9 years (standard deviation, 7.7), and 797 (49.2%) were male. Mean SFCT was 255.2 µm (standard deviation, 102.6). The normal range of SFCT was 106 to 447 µm (corresponding to 5th and 95th percentile limits of SFCT, respectively). In multivariable models, thinner SFCT was associated with older age, female gender, longer axial length, and Malay (vs. Chinese) ethnicity. In the subset of Chinese and Indian eyes (n = 1842) in whom regional variation was evaluated, the choroid was thickest at the superior and temporal sectors and thinner at the inferior and nasal sectors. CONCLUSIONS: Subfoveal choroidal thickness is influenced by age, gender, and ethnicity along with regional differences even within individual eyes. Subfoveal choroidal thickness also shows a wide range in physiologic limits. These data may be used as a reference in future studies.

Intravitreal Injection with a Conjunctival Injection Device: A Single-Center Experience.

Purpose: To evaluate the clinical performance of the intravitreal injection assistant device (InVitria) compared with the conventional freehand technique for delivering intravitreal injections. Methods: Seventy patients were randomized to receive intravitreal injections via the conventional freehand technique while 70 received injections using the InVitria. Half of all procedures in each group were performed by junior surgeons, while the rest were performed by senior surgeons. Results: Mean injections times were 90.0 ± 23.3 seconds and 64.9 ± 26.8 seconds for conventional versus InVitria (P < 0.001). Mean injection times with the conventional technique were 85.5 ± 23.0 seconds vs. 94.2 ± 23.0 seconds for senior versus junior surgeons (P = 0.120). Mean injection times with the InVitria were 56.1 ± 26.1 seconds vs. 66.3 ± 26.9 seconds (P = 0.211) for senior versus junior surgeons. There were no significant differences in pain scores regardless of technique (conventional versus In Vitria: 2.03 ± 1.73 vs. 2.13 ± 2.20, P = 0.770). Conclusions: In our experience, the InVitria is a comparable alternative to the conventional freehand technique of delivering intravitreal injections, with the potential for faster injection times and without compromising on patient comfort. Translational Relevance: The study provides evidence to suggest that the InVitria may be deployed effectively in clinical practice.

The relationship between pigment epithelial detachment and visual outcome in neovascular age-related macular degeneration and polypoidal choroidal vasculopathy.

BACKGROUND/OBJECTIVES: To compare the detailed optical coherence tomography (OCT)-based morphological parameters of pigment epithelial detachment (PED) in eyes presenting with neovascular age-related macular degeneration (nAMD) and polypoidal choroidal vasculopathy (PCV), and to assess whether these PED-associated parameters influence 1-year visual outcomes. SUBJECT/METHODS: We analysed images from a prospective observational study of treatment-naive Asian participants with nAMD or PCV. An independent reading centre graded baseline morphological features of PED on spectral-domain OCT, including greatest height, greatest width, greatest volume, morphology (predominantly dome shaped versus peaked), presence of retinal pigment epithelium (RPE) tear and cholesterol bands. The influence of these baseline features on 12 months best corrected visual acuity (BCVA) was evaluated. RESULTS: Seventy-eight eyes of 78 participants with PED were studied. In total, 40 (51.3%) participants had nAMD and 38 (48.7%) had PCV. Eyes with PCV, compared with nAMD, had PED of greater height (455.9 µm versus 389.9 µm; P = 0.035) and had higher prevalence of RPE tear (22.9 versus 5.3%; P = 0.041). In the multivariate analysis, only baseline BCVA was significantly associated with month 12 BCVA, but none of the PED-associated OCT parameters at baseline influenced month 12 BCVA. CONCLUSIONS: Despite the differences in PED height and prevalence of RPE tear between nAMD and PCV, none of these PED morphological factors on OCT at baseline significantly influenced visual outcome at 12 months.

COMPARISON OF MULTICOLOR IMAGING AND COLOR FUNDUS PHOTOGRAPHY IN THE DETECTION OF PATHOLOGICAL FINDINGS IN EYES WITH POLYPOIDAL CHOROIDAL VASCULOPATHY.

PURPOSE: To compare the appearance and frequency of detection of common features in eyes with polypoidal choroidal vasculopathy using multicolor imaging (MC) or color fundus photography (CFP). METHODS: Thirty-eight eyes with indocyanine green angiography-proven polypoidal choroidal vasculopathy, imaged with both MC and CFP, were graded by three independent retinal specialists. The presence of five prespecified pathological features (blood, exudation, polypoidal lesions, pigment epithelial detachments, and atrophy) was graded on each modality independently. Multimodal imaging including optical coherence tomography, fluorescein, and indocyanine green angiography was used as the gold standard. RESULTS: Overall, there was no statistically significant difference in the ability of MC imaging compared with CFP in detecting the pathological features. Polypoidal lesions appear as small, dark green, round lesions which have higher contrast on MC compared with a nodular orange appearance seen on CFP. Polypoidal lesions can be identified noninvasively using both MC and CFP in about half of the cases. CONCLUSION: There are differences in the appearance of polypoidal choroidal vasculopathy-associated features on MC compared with CFP. Both modalities are comparable for the detection of pathological features in eyes with polypoidal choroidal vasculopathy. MC imaging may be considered as an alternative to CFP.

Global Assessment of Retinal Arteriolar, Venular and Capillary Microcirculations Using Fundus Photographs and Optical Coherence Tomography Angiography in Diabetic Retinopathy.

Retinal arterioles, venules and capillaries are differentially affected in diabetes, and studying vascular alterations may provide information on pathogenesis of diabetic retinopathy (DR). We conducted a cross-sectional study on 49 diabetic patients, who underwent fundus photography and optical coherence tomographic angiography (OCT-A). Fundus photographs were analysed using semi-automated software for arteriolar and venular parameters, including central retinal arteriolar equivalent (CRAE), central retinal venular equivalent (CRVE) and fractal dimension (FD). Capillary parameters were measured using OCT-A, including capillary density index (CDI) and capillary FD of superficial (SVP) and deep (DVP) vascular plexuses. Severe DR was defined as severe non-proliferative DR and proliferative DR. We found that eyes with severe DR had narrower CRAE and sparser SVP CDI than eyes without. In logistic regression analysis, capillary parameters were more associated with severe DR than arteriolar or venular parameters. However, combining arteriolar, venular and capillary parameters provided the strongest association with severe DR. In linear regression analysis, eyes with poorer visual acuity had lower CRAE and FD of arterioles, venules, and DVP capillaries. We concluded that the retinal microvasculature is globally affected in severe DR, reflecting widespread microvascular impairment in perfusion. Arteriolar, venular and capillary parameters provide complementary information in assessment of DR.

Imaging in myopia: potential biomarkers, current challenges and future developments.

Myopia is rapidly increasing in Asia and around the world, while it is recognised that complications from high myopia may cause significant visual impairment. Thus, imaging the myopic eye is important for the diagnosis of sight-threatening complications, monitoring of disease progression and evaluation of treatments. For example, recent advances in high-resolution imaging using optical coherence tomography may delineate early myopic macula pathology, optical coherence tomography angiography may aid early choroidal neovascularisation detection, while multimodal imaging is important for monitoring treatment response. However, imaging the eye with high myopia accurately has its challenges and limitations, which are important for clinicians to understand in order to choose the best imaging modality and interpret the images accurately. In this review, we present the current imaging modalities available from the anterior to posterior segment of the myopic eye, including the optic nerve. We summarise the clinical indications, image interpretation and future developments that may overcome current technological limitations. We also discuss potential biomarkers for myopic progression or development of complications, including basement membrane defects, and choroidal atrophy or choroidal thickness measurements. Finally, we present future developments in the field of myopia imaging, such as photoacoustic imaging and corneal or scleral biomechanics, which may lead to innovative treatment modalities for myopia.

Efficacy and Safety of Intravitreal Aflibercept for Polypoidal Choroidal Vasculopathy in the PLANET Study: A Randomized Clinical Trial.

Importance: Polypoidal choroidal vasculopathy (PCV) is common in Asian populations, but an optimal treatment approach remains to be confirmed. Objective: To evaluate intravitreal aflibercept injection (IAI) in participants with PCV and compare IAI monotherapy with IAI plus rescue photodynamic therapy (PDT). Design, Setting, and Participants: This 96-week, double-masked, sham-controlled phase 3b/4 randomized clinical trial was conducted at multiple centers in Australia, Germany, Hong Kong, Hungary, Japan, Singapore, South Korea, and Taiwan from May 2014 to August 2016, and included adults 50 years or older with symptomatic macular PCV and a best-corrected visual acuity of 73 to 24 Early Treatment Diabetic Retinopathy Study letters (20/40-20/320 Snellen equivalent). Interventions: Participants received 2 mg of IAI at weeks 0, 4, and 8. At week 12, participants with a suboptimal response were randomized 1:1 to receive IAI plus sham PDT (IAI monotherapy) or a "rescue" of IAI plus rescue PDT (IAI/PDT). Participants who did not qualify for rescue received IAI every 8 weeks; those qualifying for rescue received IAI every 4 weeks plus sham/active PDT. When the rescue criteria were no longer met, injection intervals were gradually extended to 8 weeks. Main Outcomes and Measures: Noninferiority of IAI monotherapy to IAI/PDT for mean change in best-corrected visual acuity from baseline to week 52 (95% CI of the difference entirely above -5 letters). Results: Of the 318 participants, the mean (SD) age was 70.6 (8.2) years, 96 (30.2%) were women, and 152 (47.8%) were Japanese. Monotherapy with IAI was noninferior to IAI/PDT for the primary end point (+10.7 vs +10.8 letters, respectively; 95% CI, -2.9 to 1.6; P = .55), with few participants requiring rescue therapy (19 [12.1%] vs 23 [14.3%], respectively). Participants in both treatment groups had similar reductions in central subfield thickness from baseline to week 52 (-137.7 [IAI monotherapy] vs -143.5 µm [IAI/PDT]). At week 52, 49 (38.9%) and 60 participants (44.8%) had no polypoidal lesions observed on indocyanine green angiography in the IAI monotherapy and IAI/PDT groups, respectively. Furthermore, 116 (81.7%) and 136 (88.9%), respectively, had no polypoidal lesions with leakage. The most frequent ocular adverse events were conjunctival hemorrhage (IAI monotherapy, 8 [5.1%]) and dry eye (IAI/PDT, 9 [5.6%]). Conclusions and Relevance: Improvement in visual and/or functional outcomes was achieved in more than 85% of participants who were treated with IAI monotherapy, with no signs of leakage from polypoidal lesions in more than 80%. As fewer than 15% met the criteria of a suboptimal response to receive PDT, the potential benefit of adding PDT cannot be determined. Trial Registration: ClinicalTrials.gov Identifier: NCT02120950.

CHOROIDAL VASCULAR HYPERPERMEABILITY AS A PREDICTOR OF TREATMENT RESPONSE FOR POLYPOIDAL CHOROIDAL VASCULOPATHY.

PURPOSE: To investigate the influence of choroidal vascular hyperpermeability (CVH) and choroidal thickness on treatment outcomes in eyes with polypoidal choroidal vasculopathy (PCV) undergoing anti-vascular endothelial growth factor monotherapy or combination therapy of photodynamic therapy and anti-vascular endothelial growth factor injections. METHODS: The authors performed a prospective, observational cohort study involving 72 eyes of 72 patients with polypoidal choroidal vasculopathy (mean age 68.6 years, 51% men) treated with either monotherapy (n = 41) or combination therapy (n = 31). Each eye was imaged with color fundus photography, fluorescent angiography, indocyanine green angiography, and spectral domain optical coherence tomography. Indocyanine green angiography images were used to evaluate CVH, and spectral domain optical coherence tomography was used to measure central choroidal thickness. Changes in visual acuity over 12 months, and number of anti-vascular endothelial growth factor injections were investigated. RESULTS: Choroidal vascular hyperpermeability was present in 31 eyes (43.1%). Visual acuity change over 12 months was numerically better in the CVH group compared with the CVH (-) group (-0.099 and -0.366 logarithm of the minimal angle of resolution unit in the CVH (-) and CVH (+) groups, respectively, multivariate P = 0.063) and significantly better in a matched pair analysis (P = 0.033). Furthermore, in the combination therapy group, the number of injection was significantly lower in the CVH (+) group compared with the CVH (-) group (4.68 vs. 2.58 injections/year in the CVH (-) and CVH (+) groups; P = 0.0044). There was no significant relationship between treatment response and choroidal thickening. CONCLUSION: The presence of CVH is associated with better visual outcome in eyes with polypoidal choroidal vasculopathy and lower injection number in combination therapy. Thus, CVH, but not choroidal thickness, should be further evaluated as a potential biomarker for selecting patients for combination therapy.

Change in vision-related quality of life and influencing factors in Asians receiving treatment for neovascular age-related macular degeneration.

AIM: To assess the change in vision-related quality of life (VRQoL) after treatment for neovascular age-related macular degeneration (nAMD) and factors influencing this change in an Asian population. METHODS: In this longitudinal study, 116 patients (mean age±SD=66.5±9.9 years; 59.5% male) who underwent treatment for nAMD were recruited from a tertiary eye centre in Singapore. Best-corrected visual acuity (BCVA) and the Impact of Vision Impairment (IVI) questionnaire were evaluated at baseline and month 12. We defined three categories of BCVA change in the treated eye: BCVA gain ≥2 lines; no change in BCVA; BCVA loss ≥2 lines. The main outcome measures were the Rasch-derived IVI Reading, Mobility, and Emotional Scores. Multivariable linear regression analyses assessed the influence of sociodemographic, clinical and treatment-related factors on change in VRQoL. RESULTS: Following treatment, mean treated-eye BCVA improved by almost 2 lines (-0.22±0.40 logMAR, p<0.001) and 43% (n=50) patients reported a gain in BCVA of ≥2 lines. Mean±SD scores for Reading, Mobility and Emotional demonstrated positive changes of 0.43±1.73, 0.45±1.54 and 0.66±1.6, respectively (p<0.001 for all). In multivariable models, a ≥2 line improvement in BCVA was independently associated with a 47% (ß=0.20; CI 0.01 to 0.39) increase in Reading Scores, but was not independently associated with Mobility or Emotional Scores. CONCLUSION: Nearly half of patients undergoing treatment for nAMD reported a 2-line improvement in vision which was, in turn, associated with substantial positive increases in Reading Scores. Improvements in Mobility and Emotional Scores appear to be driven by factors other than visual acuity.

Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With Diabetes.

Importance: A deep learning system (DLS) is a machine learning technology with potential for screening diabetic retinopathy and related eye diseases. Objective: To evaluate the performance of a DLS in detecting referable diabetic retinopathy, vision-threatening diabetic retinopathy, possible glaucoma, and age-related macular degeneration (AMD) in community and clinic-based multiethnic populations with diabetes. Design, Setting, and Participants: Diagnostic performance of a DLS for diabetic retinopathy and related eye diseases was evaluated using 494 661 retinal images. A DLS was trained for detecting diabetic retinopathy (using 76 370 images), possible glaucoma (125 189 images), and AMD (72 610 images), and performance of DLS was evaluated for detecting diabetic retinopathy (using 112 648 images), possible glaucoma (71 896 images), and AMD (35 948 images). Training of the DLS was completed in May 2016, and validation of the DLS was completed in May 2017 for detection of referable diabetic retinopathy (moderate nonproliferative diabetic retinopathy or worse) and vision-threatening diabetic retinopathy (severe nonproliferative diabetic retinopathy or worse) using a primary validation data set in the Singapore National Diabetic Retinopathy Screening Program and 10 multiethnic cohorts with diabetes. Exposures: Use of a deep learning system. Main Outcomes and Measures: Area under the receiver operating characteristic curve (AUC) and sensitivity and specificity of the DLS with professional graders (retinal specialists, general ophthalmologists, trained graders, or optometrists) as the reference standard. Results: In the primary validation dataset (n = 14 880 patients; 71 896 images; mean [SD] age, 60.2 [2.2] years; 54.6% men), the prevalence of referable diabetic retinopathy was 3.0%; vision-threatening diabetic retinopathy, 0.6%; possible glaucoma, 0.1%; and AMD, 2.5%. The AUC of the DLS for referable diabetic retinopathy was 0.936 (95% CI, 0.925-0.943), sensitivity was 90.5% (95% CI, 87.3%-93.0%), and specificity was 91.6% (95% CI, 91.0%-92.2%). For vision-threatening diabetic retinopathy, AUC was 0.958 (95% CI, 0.956-0.961), sensitivity was 100% (95% CI, 94.1%-100.0%), and specificity was 91.1% (95% CI, 90.7%-91.4%). For possible glaucoma, AUC was 0.942 (95% CI, 0.929-0.954), sensitivity was 96.4% (95% CI, 81.7%-99.9%), and specificity was 87.2% (95% CI, 86.8%-87.5%). For AMD, AUC was 0.931 (95% CI, 0.928-0.935), sensitivity was 93.2% (95% CI, 91.1%-99.8%), and specificity was 88.7% (95% CI, 88.3%-89.0%). For referable diabetic retinopathy in the 10 additional datasets, AUC range was 0.889 to 0.983 (n = 40 752 images). Conclusions and Relevance: In this evaluation of retinal images from multiethnic cohorts of patients with diabetes, the DLS had high sensitivity and specificity for identifying diabetic retinopathy and related eye diseases. Further research is necessary to evaluate the applicability of the DLS in health care settings and the utility of the DLS to improve vision outcomes.

Anti-VEGF Therapy for Neovascular AMD and Polypoidal Choroidal Vasculopathy.

Anti-vascular endothelial growth factor (anti-VEGF) therapy has revolutionized the treatment of neovascular age-related macular degeneration (AMD). This review will summarize the current evidence of anti-VEGF therapy in neovascular AMD, including subtypes of retinal angiomatous proliferation and polypoidal choroidal vasculopathy (PCV). Importantly, 2 large multicenter randomized clinical trials evaluating the safety and efficacy of anti-VEGF monotherapy and combination with photodynamic therapy (PDT) have recently reported initial first-year outcomes. In this review, we summarize the latest updates in the efficacy and safety of anti-VEGF monotherapy and combination with PDT in common lesion subtypes. Remaining gaps in current understanding are highlighted where further research is needed.

Choroidal Remodeling in Age-related Macular Degeneration and Polypoidal Choroidal Vasculopathy: A 12-month Prospective Study.

Choroid thinning occurs in age-related macular degeneration (AMD). However, it remains unclear whether the reduction is due to reduction in choroidal vessels or shrinkage of choroidal stroma, or both. The purpose of this study was to evaluate the changes of the choroidal vascular and stromal area in 118 patients with typical AMD (t-AMD) and polypoidal choroidal vasculopathy (PCV) over a 12-month period. We used spectral-domain optical coherence tomography (SD-OCT) with enhanced depth imaging (EDI) mode to measure the subfoveal choroidal thickness (CT), central retinal thickness (CRT) and choroidal vascularity index (CVI - ratio of luminal area to total choroidal area). At baseline, PCV eyes had higher CRT (471.6 µm vs 439.1 µm, p = 0.02), but comparable subfoveal CT and CVI, compared to t-AMD. Eyes with high CVI at baseline showed marked reduction in stromal area compared with eyes with average or low CVI. Over 12 months, CRT and subfoveal CT significantly decreased (p < 0.001) in both subtypes. Eyes with high baseline CVI showed significant CVI reduction from baseline to month 12 (p < 0.001), whereas eyes with average to low baseline CVI showed increase in CVI. These differences in choroidal vascularity may reflect different predominant pathogenic processes and remodeling in AMD eyes with varying spectrum.

Urinary Isoprostane Levels and Age-Related Macular Degeneration.

Purpose: Oxidative stress, characterized by an excessive production of reactive oxygen intermediates has been suggested to play a role in the pathogenesis of age-related macular degeneration (AMD). We examined the association of urinary F2-isoprostanes (F2-IsoPs), a marker of lipid peroxidation and the most reliable marker of oxidative damage with AMD. Methods: We included 238 adults with AMD and 390 age- and sex-matched controls without AMD who participated in a population-based cross-sectional study in Singapore (Singapore Chinese Eye Study, 2009-2011). AMD was graded from retinal photographs using the Wisconsin Age-Related Maculopathy Grading System. Urinary-free F2-IsoPs (pmol/mmol of creatinine) were measured by gas chromatography mass spectrometry (GC-MS). The association between F2-IsoPs and AMD was examined using unconditional logistic regression models adjusted for potential confounders including smoking, body mass index (BMI), blood pressure, total and high-density lipoprotein cholesterol, and history of cardiovascular disease. Results: Higher levels of F2-IsoPs were associated with AMD independent of potential confounders. Compared to quartile 1 (Q1) of F2-IsoPs, the multivariable odds ratio (95% confidence interval) of AMD in quartiles 2, 3, and 4 were 2.05 (1.26-3.32), 1.80 (1.10-2.94), and 1.76 (1.06-2.94), respectively. In subgroup analyses comparing Q4 to Q1, this association was stronger in women, those with BMI less than 25 kg/m2 and those with hypertension, but no significant interaction was found (P interaction > 0.1 for each strata). Conclusions: Higher levels of urinary F2-IsoPs levels were associated with AMD independent of potential confounders in Chinese adults.

Prevalence and risk factors for epiretinal membrane: the Singapore Epidemiology of Eye Disease study.

AIM: To examine prevalence and risk factors of epiretinal membrane (ERM) in a large, contemporary, multiethnic Asian population. METHODS: Combined analysis of three population-based studies of eye diseases, with a total of 9799 Chinese, Malays and Indians residing in the general communities of Singapore. A comprehensive ophthalmic examination, interviews and laboratory blood tests were performed to assess potential risk factors. Digital retinal photographs were used to assess ERM according to a standardised protocol. ERM was classified into cellophane macular reflex (CMR) and/or preretinal macular fibrosis (PMF), and also as primary or secondary (in eyes with other retinal pathology or a history of cataract surgery). RESULTS: The age-standardised and ethnicity-standardised prevalence was 12.1% for any ERM, 6.8% for CMR, 6.7% for PMF and 2.8% for bilateral ERM. ERM prevalence was higher in Chinese (13.0%) compared with Malays (7.9%) or Indians (8.7%). In multivariate analysis, significant factors associated with primary ERM were older age (OR 1.08 per year increase; p<0.01), Chinese ethnicity (OR 1.60 vs Indians; p<0.01; OR 1.39 vs Malays; p<0.01), smoking (OR 0.70; p=0.01), longer axial length (OR 1.07 per mm increase; p=0.03) and cataract (OR 0.64; p<0.01). Significant factors independently associated with secondary ERM were older age (OR 1.05; p<0.01), cataract surgery (OR 10.6; p<0.01) and diabetic retinopathy (OR 2.48; p<0.01). CONCLUSIONS: ERM is common in Asians, particularly among Chinese. Older age is the most consistent risk factor for any ERM, and previous cataract surgery and diabetic retinopathy are the strongest risk factors for secondary ERM.

Choroidal Structural Changes in Myopic Choroidal Neovascularization After Treatment With Antivascular Endothelial Growth Factor Over 1 Year.

PURPOSE: To evaluate choroidal structural changes in eyes with myopic choroidal neovascularization (mCNV) treated with anti-VEGF over 12 months. METHODS: We prospectively evaluated subfoveal choroidal thickness (SFCT) and choroidal vascularity index (CVI) using spectral-domain optical coherence tomography (SD-OCT) at baseline, 6, and 12 months in both eyes in patients presenting with unilateral mCNV. Choroidal vascularity index was defined as the ratio of luminal area to total choroidal area after SD-OCT images were binarized digitally. RESULTS: We included 20 patients (20 eyes with mCNV and 20 fellow eyes without mCNV) with mean age of 60.35 ± 10.85 years. At baseline, mean SFCT and CVI was similar between eyes with mCNV and fellow eyes (69.20 ± 63.04 µm vs. 67.10 ± 65.74 µm, P = 0.713 for SFCT and 59.44 ± 3.92% vs. 59.03 ±. 5.58%, P = 0.958 for CVI). Subfoveal choroidal thickness decreased significantly in the mCNV eyes to 54.75 ± 45.43 µm (P = 0.017) at 12 months after anti-VEGF therapy, whereas SFCT in the contralateral eyes did not change significantly. There was no significant change in CVI in mCNV eyes or contralateral eyes from baseline to 12 months. Thinning of SFCT did not influence final BCVA. CONCLUSIONS: Thinning of subfoveal choroid without alteration in CVI was observed in eyes with mCNV treated with anti-VEGF therapy over 12 months. This finding may be explained by mechanical stretching in response to globe expansion.