Extending the diabetic retinopathy screening intervals in Singapore: methodology and preliminary findings of a cohort study.

BACKGROUND: The Diabetic Retinopathy Extended Screening Study (DRESS) aims to develop and validate a new DR/diabetic macular edema (DME) risk stratification model in patients with Type 2 diabetes (DM) to identify low-risk groups who can be safely assigned to biennial or triennial screening intervals. We describe the study methodology, participants' baseline characteristics, and preliminary DR progression rates at the first annual follow-up. METHODS: DRESS is a 3-year ongoing longitudinal study of patients with T2DM and no or mild non-proliferative DR (NPDR, non-referable) who underwent teleophthalmic screening under the Singapore integrated Diabetic Retinopathy Programme (SiDRP) at four SingHealth Polyclinics. Patients with referable DR/DME (> mild NPDR) or ungradable fundus images were excluded. Sociodemographic, lifestyle, medical and clinical information was obtained from medical records and interviewer-administered questionnaires at baseline. These data are extracted from medical records at 12, 24 and 36 months post-enrollment. Baseline descriptive characteristics stratified by DR severity at baseline and rates of progression to referable DR at 12-month follow-up were calculated. RESULTS: Of 5,840 eligible patients, 78.3% (n = 4,570, median [interquartile range [IQR] age 61.0 [55-67] years; 54.7% male; 68.0% Chinese) completed the baseline assessment. At baseline, 97.4% and 2.6% had none and mild NPDR (worse eye), respectively. Most participants had hypertension (79.2%) and dyslipidemia (92.8%); and almost half were obese (43.4%, BMI ≥ 27.5 kg/m2). Participants without DR (vs mild DR) reported shorter DM duration, and had lower haemoglobin A1c, triglycerides and urine albumin/creatinine ratio (all p < 0.05). To date, we have extracted 41.8% (n = 1909) of the 12-month follow-up data. Of these, 99.7% (n = 1,904) did not progress to referable DR. Those who progressed to referable DR status (0.3%) had no DR at baseline. CONCLUSIONS: In our prospective study of patients with T2DM and non-referable DR attending polyclinics, we found extremely low annual DR progression rates. These preliminary results suggest that extending screening intervals beyond 12 months may be viable and safe for most participants, although our 3-year follow up data are needed to substantiate this claim and develop the risk stratification model to identify low-risk patients with T2DM who can be assigned biennial or triennial screening intervals.

A deep learning system for predicting time to progression of diabetic retinopathy.

Diabetic retinopathy (DR) is the leading cause of preventable blindness worldwide. The risk of DR progression is highly variable among different individuals, making it difficult to predict risk and personalize screening intervals. We developed and validated a deep learning system (DeepDR Plus) to predict time to DR progression within 5 years solely from fundus images. First, we used 717,308 fundus images from 179,327 participants with diabetes to pretrain the system. Subsequently, we trained and validated the system with a multiethnic dataset comprising 118,868 images from 29,868 participants with diabetes. For predicting time to DR progression, the system achieved concordance indexes of 0.754-0.846 and integrated Brier scores of 0.153-0.241 for all times up to 5 years. Furthermore, we validated the system in real-world cohorts of participants with diabetes. The integration with clinical workflow could potentially extend the mean screening interval from 12 months to 31.97 months, and the percentage of participants recommended to be screened at 1-5 years was 30.62%, 20.00%, 19.63%, 11.85% and 17.89%, respectively, while delayed detection of progression to vision-threatening DR was 0.18%. Altogether, the DeepDR Plus system could predict individualized risk and time to DR progression over 5 years, potentially allowing personalized screening intervals.

Evaluating the outcome of screening for glaucoma using colour fundus photography-based referral criteria in a teleophthalmology screening programme for diabetic retinopathy.

AIMS: To evaluate the effectiveness of glaucoma screening using glaucoma suspect (GS) referral criteria assessed on colour fundus photographs in Singapore's Integrated Diabetic Retinopathy Programme (SiDRP). METHODS: A case-control study. This study included diabetic subjects who were referred from SiDRP with and without GS between January 2017 and December 2018 and reviewed at Singapore National Eye Centre. The GS referral criteria were based on the presence of a vertical cup-to-disc ratio (VCDR) of ≥0.65 and other GS features. The final glaucoma diagnosis confirmed from electronic medical records was retrospectively matched with GS status. The sensitivity, specificity and positive predictive value (PPV) of the test were evaluated. RESULTS: Of 5023 patients (2625 with GS and 2398 without GS) reviewed for glaucoma, 451 (9.0%, 95% CI 8.2% to 9.8%) were confirmed as glaucoma. The average follow-up time was 21.5±10.2 months. Using our current GS referral criteria, the sensitivity, specificity and PPV were 81.6% (95% CI 77.7% to 85.1%), 50.6% (95% CI 49.2% to 52.1%) and 14.0% (95% CI 13.4% to 14.7%), respectively, resulting in 2257 false positive cases. Increasing the VCDR cut-off for referral to ≥0.80, the specificity increased to 93.9% (95% CI 93.1% to 94.5%) but the sensitivity decreased to 11.3% (95% CI 8.5% to 14.6%), with a PPV of 15.4% (95% CI 12.0% to 19.4%). CONCLUSIONS: Opportunistic screening for glaucoma in a lower VCDR group could result in a high number of unnecessary referrals. If healthcare infrastructures are limited, targeting case findings on a larger VCDR group with high specificity will still be beneficial.

Deep learning algorithms to detect diabetic kidney disease from retinal photographs in multiethnic populations with diabetes.

OBJECTIVE: To develop a deep learning algorithm (DLA) to detect diabetic kideny disease (DKD) from retinal photographs of patients with diabetes, and evaluate performance in multiethnic populations. MATERIALS AND METHODS: We trained 3 models: (1) image-only; (2) risk factor (RF)-only multivariable logistic regression (LR) model adjusted for age, sex, ethnicity, diabetes duration, HbA1c, systolic blood pressure; (3) hybrid multivariable LR model combining RF data and standardized z-scores from image-only model. Data from Singapore Integrated Diabetic Retinopathy Program (SiDRP) were used to develop (6066 participants with diabetes, primary-care-based) and internally validate (5-fold cross-validation) the models. External testing on 2 independent datasets: (1) Singapore Epidemiology of Eye Diseases (SEED) study (1885 participants with diabetes, population-based); (2) Singapore Macroangiopathy and Microvascular Reactivity in Type 2 Diabetes (SMART2D) (439 participants with diabetes, cross-sectional) in Singapore. Supplementary external testing on 2 Caucasian cohorts: (3) Australian Eye and Heart Study (AHES) (460 participants with diabetes, cross-sectional) and (4) Northern Ireland Cohort for the Longitudinal Study of Ageing (NICOLA) (265 participants with diabetes, cross-sectional). RESULTS: In SiDRP validation, area under the curve (AUC) was 0.826(95% CI 0.818-0.833) for image-only, 0.847(0.840-0.854) for RF-only, and 0.866(0.859-0.872) for hybrid. Estimates with SEED were 0.764(0.743-0.785) for image-only, 0.802(0.783-0.822) for RF-only, and 0.828(0.810-0.846) for hybrid. In SMART2D, AUC was 0.726(0.686-0.765) for image-only, 0.701(0.660-0.741) in RF-only, 0.761(0.724-0.797) for hybrid. DISCUSSION AND CONCLUSION: There is potential for DLA using retinal images as a screening adjunct for DKD among individuals with diabetes. This can value-add to existing DLA systems which diagnose diabetic retinopathy from retinal images, facilitating primary screening for DKD.

Detecting visually significant cataract using retinal photograph-based deep learning.

Age-related cataracts are the leading cause of visual impairment among older adults. Many significant cases remain undiagnosed or neglected in communities, due to limited availability or accessibility to cataract screening. In the present study, we report the development and validation of a retinal photograph-based, deep-learning algorithm for automated detection of visually significant cataracts, using more than 25,000 images from population-based studies. In the internal test set, the area under the receiver operating characteristic curve (AUROC) was 96.6%. External testing performed across three studies showed AUROCs of 91.6-96.5%. In a separate test set of 186 eyes, we further compared the algorithm's performance with 4 ophthalmologists' evaluations. The algorithm performed comparably, if not being slightly more superior (sensitivity of 93.3% versus 51.7-96.6% by ophthalmologists and specificity of 99.0% versus 90.7-97.9% by ophthalmologists). Our findings show the potential of a retinal photograph-based screening tool for visually significant cataracts among older adults, providing more appropriate referrals to tertiary eye centers.

Macular neovascularization in eyes with pachydrusen.

The natural history and clinical significance of pachydrusen is unclear. This study aims to compare the longitudinal changes of eyes with pachydrusen and soft drusen and progression to exudative macular neovascularisation (MNV). Patients with a diagnosis of MNV in one eye only and the fellow eye was selected as the study eye. Study eyes were required to have pachydrusen or soft drusen on fundus photographs and follow up of at least 2 years or until exudative MNV occurred. Systematic grading was performed at baseline and change in drusen area and onset of exudative MNV recorded over the period of follow up. A total of 75 eyes from 75 patients (29 with pachydrusen and 46 with soft drusen) were included. There was no difference in the rate of progression to exudative MNV in the soft and pachydrusen groups (13.3% versus 24.1%, p = 0.38). Pachydrusen, as compared to soft drusen, was associated with polypoidal choroidal vasculopathy subtype (85.7% versus 16.7%, p < 0.01) and the location of exudation was co-localised with soft drusen but not with pachydrusen. There was a higher rate of increase in soft drusen area compared to pachydrusen area (27.7 ± 31.9%/year versus 8.7 ± 12.4%/year respectively, p < 0.01). We found no difference in the proportion of eyes that developed exudative MNV in this study however characterisation of drusen evolution patterns revealed a strong association with exudative MNV subtype.

Referral for disease-related visual impairment using retinal photograph-based deep learning: a proof-of-concept, model development study.

BACKGROUND: In current approaches to vision screening in the community, a simple and efficient process is needed to identify individuals who should be referred to tertiary eye care centres for vision loss related to eye diseases. The emergence of deep learning technology offers new opportunities to revolutionise this clinical referral pathway. We aimed to assess the performance of a newly developed deep learning algorithm for detection of disease-related visual impairment. METHODS: In this proof-of-concept study, using retinal fundus images from 15 175 eyes with complete data related to best-corrected visual acuity or pinhole visual acuity from the Singapore Epidemiology of Eye Diseases Study, we first developed a single-modality deep learning algorithm based on retinal photographs alone for detection of any disease-related visual impairment (defined as eyes from patients with major eye diseases and best-corrected visual acuity of <20/40), and moderate or worse disease-related visual impairment (eyes with disease and best-corrected visual acuity of <20/60). After development of the algorithm, we tested it internally, using a new set of 3803 eyes from the Singapore Epidemiology of Eye Diseases Study. We then tested it externally using three population-based studies (the Beijing Eye study [6239 eyes], Central India Eye and Medical study [6526 eyes], and Blue Mountains Eye Study [2002 eyes]), and two clinical studies (the Chinese University of Hong Kong's Sight Threatening Diabetic Retinopathy study [971 eyes] and the Outram Polyclinic Study [1225 eyes]). The algorithm's performance in each dataset was assessed on the basis of the area under the receiver operating characteristic curve (AUC). FINDINGS: In the internal test dataset, the AUC for detection of any disease-related visual impairment was 94·2% (95% CI 93·0-95·3; sensitivity 90·7% [87·0-93·6]; specificity 86·8% [85·6-87·9]). The AUC for moderate or worse disease-related visual impairment was 93·9% (95% CI 92·2-95·6; sensitivity 94·6% [89·6-97·6]; specificity 81·3% [80·0-82·5]). Across the five external test datasets (16 993 eyes), the algorithm achieved AUCs ranging between 86·6% (83·4-89·7; sensitivity 87·5% [80·7-92·5]; specificity 70·0% [66·7-73·1]) and 93·6% (92·4-94·8; sensitivity 87·8% [84·1-90·9]; specificity 87·1% [86·2-88·0]) for any disease-related visual impairment, and the AUCs for moderate or worse disease-related visual impairment ranged between 85·9% (81·8-90·1; sensitivity 84·7% [73·0-92·8]; specificity 74·4% [71·4-77·2]) and 93·5% (91·7-95·3; sensitivity 90·3% [84·2-94·6]; specificity 84·2% [83·2-85·1]). INTERPRETATION: This proof-of-concept study shows the potential of a single-modality, function-focused tool in identifying visual impairment related to major eye diseases, providing more timely and pinpointed referral of patients with disease-related visual impairment from the community to tertiary eye hospitals. FUNDING: National Medical Research Council, Singapore.

Artificial intelligence for teleophthalmology-based diabetic retinopathy screening in a national programme: an economic analysis modelling study.

BACKGROUND: Deep learning is a novel machine learning technique that has been shown to be as effective as human graders in detecting diabetic retinopathy from fundus photographs. We used a cost-minimisation analysis to evaluate the potential savings of two deep learning approaches as compared with the current human assessment: a semi-automated deep learning model as a triage filter before secondary human assessment; and a fully automated deep learning model without human assessment. METHODS: In this economic analysis modelling study, using 39 006 consecutive patients with diabetes in a national diabetic retinopathy screening programme in Singapore in 2015, we used a decision tree model and TreeAge Pro to compare the actual cost of screening this cohort with human graders against the simulated cost for semi-automated and fully automated screening models. Model parameters included diabetic retinopathy prevalence rates, diabetic retinopathy screening costs under each screening model, cost of medical consultation, and diagnostic performance (ie, sensitivity and specificity). The primary outcome was total cost for each screening model. Deterministic sensitivity analyses were done to gauge the sensitivity of the results to key model assumptions. FINDINGS: From the health system perspective, the semi-automated screening model was the least expensive of the three models, at US$62 per patient per year. The fully automated model was $66 per patient per year, and the human assessment model was $77 per patient per year. The savings to the Singapore health system associated with switching to the semi-automated model are estimated to be $489 000, which is roughly 20% of the current annual screening cost. By 2050, Singapore is projected to have 1 million people with diabetes; at this time, the estimated annual savings would be $15 million. INTERPRETATION: This study provides a strong economic rationale for using deep learning systems as an assistive tool to screen for diabetic retinopathy. FUNDING: Ministry of Health, Singapore.

A deep learning algorithm to detect chronic kidney disease from retinal photographs in community-based populations.

BACKGROUND: Screening for chronic kidney disease is a challenge in community and primary care settings, even in high-income countries. We developed an artificial intelligence deep learning algorithm (DLA) to detect chronic kidney disease from retinal images, which could add to existing chronic kidney disease screening strategies. METHODS: We used data from three population-based, multiethnic, cross-sectional studies in Singapore and China. The Singapore Epidemiology of Eye Diseases study (SEED, patients aged ≥40 years) was used to develop (5188 patients) and validate (1297 patients) the DLA. External testing was done on two independent datasets: the Singapore Prospective Study Program (SP2, 3735 patients aged ≥25 years) and the Beijing Eye Study (BES, 1538 patients aged ≥40 years). Chronic kidney disease was defined as estimated glomerular filtration rate less than 60 mL/min per 1·73m2. Three models were trained: 1) image DLA; 2) risk factors (RF) including age, sex, ethnicity, diabetes, and hypertension; and 3) hybrid DLA combining image and RF. Model performances were evaluated using the area under the receiver operating characteristic curve (AUC). FINDINGS: In the SEED validation dataset, the AUC was 0·911 for image DLA (95% CI 0·886 -0·936), 0·916 for RF (0·891-0·941), and 0·938 for hybrid DLA (0·917-0·959). Corresponding estimates in the SP2 testing dataset were 0·733 for image DLA (95% CI 0·696-0·770), 0·829 for RF (0·797-0·861), and 0·810 for hybrid DLA (0·776-0·844); and in the BES testing dataset estimates were 0·835 for image DLA (0·767-0·903), 0·887 for RF (0·828-0·946), and 0·858 for hybrid DLA (0·794-0·922). AUC estimates were similar in subgroups of people with diabetes (image DLA 0·889 [95% CI 0·850-0·928], RF 0·899 [0·862-0·936], hybrid 0·925 [0·893-0·957]) and hypertension (image DLA 0·889 [95% CI 0·860-0·918], RF 0·889 [0·860-0·918], hybrid 0·918 [0·893-0·943]). INTERPRETATION: A retinal image DLA shows good performance for estimating chronic kidney disease, underlying the feasibility of using retinal photography as an adjunctive or opportunistic screening tool for chronic kidney disease in community populations. FUNDING: National Medical Research Council, Singapore.

Exploring Factors Underlying Ethnic Difference in Age-related Macular Degeneration Prevalence.

AIMS: To assess contributions of dietary and genetic factors to ethnic differences in AMD prevalence. DESIGN: Population-based analytical study. METHODS: In the Blue Mountains Eye Study, Australia (European ancestry n = 2826) and Multi-Ethnic Cohort Study, Singapore (Asian ancestry, n = 1900), AMD was assessed from retinal photographs. Patterns of dietary composition and scores of the Alternative Healthy Eating Index were computed using food frequency questionnaire data. Genetic susceptibility to AMD was determined using either single nucleotide polymorphisms (SNPs) of the complement factor H and age-related maculopathy susceptibility 2 genes, or combined odds-weighted genetic risk scores of 24 AMD-associated SNPs. Associations of AMD with ethnicity, diet, and genetics were assessed using logistic regression. Six potential mediators covering genetic, diet and lifestyle factors were assessed for their contributions to AMD risk difference between the two samples using mediation analyses. RESULTS: Age-standardized prevalence of any (early or late) AMD was higher in the European (16%) compared to Asian samples (9%, p < .01). Mean AMD-related genetic risk scores were also higher in European (33.3 ± 4.4) than Asian (Chinese) samples (31.7 ± 3.7, p < .001). In a model simultaneously adjusting for age, ethnicity, genetic susceptibility and Alternative Healthy Eating Index scores, only age and genetic susceptibility were significantly associated with AMD. Genetic risk scores contributed 19% of AMD risk difference between the two samples while intake of polyunsaturated fatty acids contributed 7.2%. CONCLUSION: Genetic susceptibility to AMD was higher in European compared to Chinese samples and explained more of the AMD risk difference between the two samples than the dietary factors investigated.

Technical and imaging factors influencing performance of deep learning systems for diabetic retinopathy.

Deep learning (DL) has been shown to be effective in developing diabetic retinopathy (DR) algorithms, possibly tackling financial and manpower challenges hindering implementation of DR screening. However, our systematic review of the literature reveals few studies studied the impact of different factors on these DL algorithms, that are important for clinical deployment in real-world settings. Using 455,491 retinal images, we evaluated two technical and three image-related factors in detection of referable DR. For technical factors, the performances of four DL models (VGGNet, ResNet, DenseNet, Ensemble) and two computational frameworks (Caffe, TensorFlow) were evaluated while for image-related factors, we evaluated image compression levels (reducing image size, 350, 300, 250, 200, 150 KB), number of fields (7-field, 2-field, 1-field) and media clarity (pseudophakic vs phakic). In detection of referable DR, four DL models showed comparable diagnostic performance (AUC 0.936-0.944). To develop the VGGNet model, two computational frameworks had similar AUC (0.936). The DL performance dropped when image size decreased below 250 KB (AUC 0.936, 0.900, p < 0.001). The DL performance performed better when there were increased number of fields (dataset 1: 2-field vs 1-field-AUC 0.936 vs 0.908, p < 0.001; dataset 2: 7-field vs 2-field vs 1-field, AUC 0.949 vs 0.911 vs 0.895). DL performed better in the pseudophakic than phakic eyes (AUC 0.918 vs 0.833, p < 0.001). Various image-related factors play more significant roles than technical factors in determining the diagnostic performance, suggesting the importance of having robust training and testing datasets for DL training and deployment in the real-world settings.

Artificial intelligence using deep learning to screen for referable and vision-threatening diabetic retinopathy in Africa: a clinical validation study.

BACKGROUND: Radical measures are required to identify and reduce blindness due to diabetes to achieve the Sustainable Development Goals by 2030. Therefore, we evaluated the accuracy of an artificial intelligence (AI) model using deep learning in a population-based diabetic retinopathy screening programme in Zambia, a lower-middle-income country. METHODS: We adopted an ensemble AI model consisting of a combination of two convolutional neural networks (an adapted VGGNet architecture and a residual neural network architecture) for classifying retinal colour fundus images. We trained our model on 76 370 retinal fundus images from 13 099 patients with diabetes who had participated in the Singapore Integrated Diabetic Retinopathy Program, between 2010 and 2013, which has been published previously. In this clinical validation study, we included all patients with a diagnosis of diabetes that attended a mobile screening unit in five urban centres in the Copperbelt province of Zambia from Feb 1 to June 31, 2012. In our model, referable diabetic retinopathy was defined as moderate non-proliferative diabetic retinopathy or worse, diabetic macular oedema, and ungradable images. Vision-threatening diabetic retinopathy comprised severe non-proliferative and proliferative diabetic retinopathy. We calculated the area under the curve (AUC), sensitivity, and specificity for referable diabetic retinopathy, and sensitivities of vision-threatening diabetic retinopathy and diabetic macular oedema compared with the grading by retinal specialists. We did a multivariate analysis for systemic risk factors and referable diabetic retinopathy between AI and human graders. FINDINGS: A total of 4504 retinal fundus images from 3093 eyes of 1574 Zambians with diabetes were prospectively recruited. Referable diabetic retinopathy was found in 697 (22·5%) eyes, vision-threatening diabetic retinopathy in 171 (5·5%) eyes, and diabetic macular oedema in 249 (8·1%) eyes. The AUC of the AI system for referable diabetic retinopathy was 0·973 (95% CI 0·969-0·978), with corresponding sensitivity of 92·25% (90·10-94·12) and specificity of 89·04% (87·85-90·28). Vision-threatening diabetic retinopathy sensitivity was 99·42% (99·15-99·68) and diabetic macular oedema sensitivity was 97·19% (96·61-97·77). The AI model and human graders showed similar outcomes in referable diabetic retinopathy prevalence detection and systemic risk factors associations. Both the AI model and human graders identified longer duration of diabetes, higher level of glycated haemoglobin, and increased systolic blood pressure as risk factors associated with referable diabetic retinopathy. INTERPRETATION: An AI system shows clinically acceptable performance in detecting referable diabetic retinopathy, vision-threatening diabetic retinopathy, and diabetic macular oedema in population-based diabetic retinopathy screening. This shows the potential application and adoption of such AI technology in an under-resourced African population to reduce the incidence of preventable blindness, even when the model is trained in a different population. FUNDING: National Medical Research Council Health Service Research Grant, Large Collaborative Grant, Ministry of Health, Singapore; the SingHealth Foundation; and the Tanoto Foundation.

Association of serum lutein and zeaxanthin with quantitative measures of retinal vascular parameters.

To evaluate the association between serum carotenoids and quantitative measures of retinal vasculature in elderly Singapore Chinese subjects. The following details were collected in 128 healthy subjects: sociodemographics, lifestyle information, medical and drug history, and anthropometric measurements. Serum concentrations of carotenoids were estimated in fasting venous blood using high performance liquid chromatography. Retinal vascular parameters were quantitatively measured from retinal photographs using a computer-assisted program (Singapore I Vessel Assessment). The mean age of the population was 54.1 years (range 40 to 81 years). In multiple linear regression analysis, per SD decrease in retinal arteriolar caliber [ß = 0.045 (0.003 to 0.086), p = 0.036], per SD increase in retinal venular caliber [ß = -0.045 (-0.086 to -0.003), p = 0.036] and per SD increase in arteriolar branching angle [ß = -0.039 (-0.072 to -0.006), p = 0.021] were associated with decreased serum lutein. Per SD increase in retinal venular tortuosity [ß = -0.0075 (-0.0145 to -0.0004), p = 0.039] and per SD increase in arteriolar branching angle (ß = -0.0073 [-0.0142 to -0.0059], p = 0.041) were associated with decreased serum zeaxanthin. None of the other carotenoids demonstrated meaningful relationship with quantitative measures of retinal vasculature. Lower levels of lutein and zeaxanthin demonstrated significant relationship with adverse quantitative measures of retinal vasculature in elderly healthy subjects.

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.

Availability and variability in guidelines on diabetic retinopathy screening in Asian countries.

BACKGROUND: Diabetic retinopathy (DR) is a blinding yet treatable complication of diabetes. DR screening is highly cost-effective at reducing blindness. Amidst the rapidly growing diabetic population in Asia, the prevalence of DR in the region is relatively less well known. AIMS: To review existing national DR screening guidelines of 50 countries in Asia, compare them against the International Council of Ophthalmology (ICO) guideline, and summarise the prevalence rates of DR and sight-threatening DR (STDR) in these countries. METHODS: We systematically searched for published guidelines from the National Guideline Clearinghouse and other databases, and contacted local diabetic and ophthalmological associations of all 50 Asian countries. RESULTS: Eleven Asian countries have published relevant guidelines, nine of which pertain to general diabetes care and two are DR-specific, covering less than half of Asia's population. The median DR prevalence among patients with diabetes is 30.5% (IQR: 23.2%-36.8%), similar to the USA and the UK. However, rates of STDR are consistently higher. All guidelines from the 11 Asian countries fulfil the ICO standard on when to start and repeat screening, except for screening interval for pregnant patients. However, only 2 of the 11 guidelines fulfil the ICO referral criteria and 6 partially fulfil. A third of the recommendations on screening process, equipment and personnel is either unavailable or incomplete. CONCLUSIONS: Countries in Asia need to establish more comprehensive and evidence-based DR screening guidelines to facilitate the execution of robust screening programmes that could help reduce DR-related blindness, improve patient outcomes and reduce healthcare costs.

Retinal vascular caliber between eyes with asymmetric glaucoma.

PURPOSE: To compare differences in retinal arterial and venular caliber (RAC and RVC respectively) between fellow eyes with glaucoma of asymmetric severity. METHODS: We included subjects with bilateral primary glaucoma that had vertical cup-disc ratios (VCDR) >0.2 between both eyes, or visual field (VF) mean deviation (MD) >6.0 decibels (dB) between both eyes. RESULTS: Among 158 subjects, the average RAC in glaucoma eyes was 131.5 ± 17.8 µm vs 141.6 ± 18.8 µm in fellow eyes with mild disease (p < 0.001). RVCs in glaucoma eyes were 201.0 ± 21.4 µm vs 211.7 ± 25.3 µm in fellow eyes with mild disease (p < 0.001). This relationship held in clustered linear regression models adjusted for age, gender, vascular risk factors, visual acuity, axial length, and intraocular pressure, with RVCs narrower in eyes with worse disease vs mild disease. Eyes with worse disease had greater VCDR (0.9 ± 0.1 vs 0.7 ± 0.1, p < 0.001), and worse VF MD (-18.5 ± 8.6 vs -6.6 ± 5.6, p < 0.001). CONCLUSION: In glaucoma with asymmetric severity between fellow eyes, retinal vascular caliber is less in the eye with more severe disease.

Myopia-related fundus changes in Singapore adults with high myopia.

PURPOSE: To examine the pattern of myopia-related macular and optic disc changes in Singapore adults with high myopia (spherical equivalent ≤-6.00 diopters). DESIGN: Asian adults with high myopia from 3 population-based surveys. METHODS: Adults 40 years and older (n = 359) with high myopia were pooled from 3 population-based surveys in Singapore Asians: (1) the Singapore Prospective Study Program (SP2, n = 184); (2) the Singapore Malay Eye Study (SiMES, n = 98); and (3) the Singapore Indian Eye Study (SINDI, n = 77). All study participants underwent standardized refraction and fundus photography, and SiMES and SINDI subjects also completed ocular biometry measurements. Myopia-related macular (posterior staphyloma, lacquer cracks, Fuchs spot, myopic chorioretinal atrophy, and myopic choroidal neovascularization) and optic disc (optic nerve head tilt, optic disc dimensions, and peripapillary atrophy) changes were evaluated. RESULTS: The most common myopia-related macular finding in adults with high myopia was staphyloma (23%), followed by chorioretinal atrophy (19.3%). There were few cases of lacquer crack (n = 6, 1.8%), T-sign (n = 6, 1.8%), retinal hemorrhage (n = 3, 0.9%), active myopic choroidal neovascularization (n = 3, 0.9%), and no case of Fuchs spot. The most common disc finding associated with high myopia was peripapillary atrophy (81.2%), followed by disc tilt (57.4%). Staphyloma and chorioretinal atrophy increased in prevalence with increasing age, increasing myopic refractive error, and increasing axial length (all P < .001). Ethnicity comparisons demonstrated the highest proportion of staphyloma (P = .04) among Malays, the highest proportion of peripapillary atrophy (P = .01) and disc tilt (P < .001) among Chinese, and the largest cup-to-disc ratio (P < .001) among Indians. CONCLUSIONS: Staphyloma and chorioretinal atrophy lesions were the most common fundus findings among Asian adults with high myopia. In this population, tilted discs and peripapillary atrophy were also common, while choroidal neovascularization and Fuchs spot were rare. In contrast with Singapore teenagers, in whom tilted disc and peripapillary atrophy were common while staphyloma and chorioretinal atrophy were rare, pathologic myopia appears to be dependent on the duration of disease and, thus, age of the individual.

Accuracy of diabetic retinopathy screening by trained non-physician graders using non-mydriatic fundus camera.

INTRODUCTION: We compared the agreement of diabetic retinopathy (DR) assessment between trained non-physician graders (NPGs) and family physicians (FPs) in a primary healthcare setting. METHODS: This was a cross-sectional study conducted retrospectively over a period of one month. The participants were diabetic patients from two primary healthcare clinics (polyclinics) in Singapore. Single-field digital retinal images were obtained using a non-mydriatic 45-degree fundus camera. Retinal images were graded for the presence or absence of DR by FPs at the polyclinics and by NPGs at a central ocular grading centre. The FPs' and NPGs' assessments of DR were compared with readings by a single retinal specialist (reference standard). RESULTS: A total of 367 diabetic patients (706 eyes) were included in the study. The mean age of the patients was 63 years, and the majority were Chinese (83.8%). For DR assessment, the agreement between NPGs and the retinal specialist was substantial (ĸ = 0.66), while the agreement between FPs and the retinal specialist was only fair (ĸ = 0.40). NPGs' assessment showed higher sensitivity (70% vs. 45%) and comparable specificity (94% vs. 92%) as compared to FPs' assessment. The area under the receiver operating characteristic curve of NPGs' assessment of DR was greater than that of the FPs' (0.82 vs. 0.69, p < 0.001). CONCLUSION: This study has demonstrated that trained NPGs are able to provide good detection of DR and maculopathy from fundus photographs. Our findings suggest that DR screening by trained NPGs may provide a costeffective alternative to FPs.