Using deep learning to detect diabetic retinopathy on handheld non-mydriatic retinal images acquired by field workers in community settings.

Diabetic retinopathy (DR) at risk of vision loss (referable DR) needs to be identified by retinal screening and referred to an ophthalmologist. Existing automated algorithms have mostly been developed from images acquired with high cost mydriatic retinal cameras and cannot be applied in the settings used in most low- and middle-income countries. In this prospective multicentre study, we developed a deep learning system (DLS) that detects referable DR from retinal images acquired using handheld non-mydriatic fundus camera by non-technical field workers in 20 sites across India. Macula-centred and optic-disc-centred images from 16,247 eyes (9778 participants) were used to train and cross-validate the DLS and risk factor based logistic regression models. The DLS achieved an AUROC of 0.99 (1000 times bootstrapped 95% CI 0.98-0.99) using two-field retinal images, with 93.86 (91.34-96.08) sensitivity and 96.00 (94.68-98.09) specificity at the Youden's index operational point. With single field inputs, the DLS reached AUROC of 0.98 (0.98-0.98) for the macula field and 0.96 (0.95-0.98) for the optic-disc field. Intergrader performance was 90.01 (88.95-91.01) sensitivity and 96.09 (95.72-96.42) specificity. The image based DLS outperformed all risk factor-based models. This DLS demonstrated a clinically acceptable performance for the identification of referable DR despite challenging image capture conditions.

Validation of diagnostic accuracy of retinal image grading by trained non-ophthalmologist grader for detecting diabetic retinopathy and diabetic macular edema.

PURPOSE: To validate the fundus image grading results by a trained grader (Non-ophthalmologist) and an ophthalmologist grader for detecting diabetic retinopathy (DR) and diabetic macular oedema (DMO) against fundus examination by a retina specialist (gold standard). METHODS: A prospective diagnostic accuracy study was conducted using 2002 non-mydriatic colour fundus images from 1001 patients aged ≥40 years. Using the Aravind Diabetic Retinopathy Evaluation Software (ADRES) images were graded by both a trained non-ophthalmologist grader (grader-1) and an ophthalmologist (grader-2). Sensitivity, specificity, positive predictive value and negative predictive value were calculated for grader-1 and grader-2 against the grading results by an independent retina specialist who performed dilated fundus examination for every study participant. RESULTS: Out of 1001 patients included, 42% were women and the mean ± (SD) age was 55.8 (8.39) years. For moderate or worse DR, the sensitivity and specificity for grading by grader-1 with respect to the gold standard was 66.9% and 91.0% respectively and the same for the ophthalmologist was 83.6% and 80.3% respectively. For referable DMO, grader-1 and grader-2 had a sensitivity of 74.6% and 85.6% respectively and a specificity of 83.7% and 79.8% respectively. CONCLUSIONS: Our results demonstrate good level of accuracy for the fundus image grading performed by a trained non-ophthalmologist which was comparable with the grading by an ophthalmologist. Engaging trained non-ophthalmologists potentially can enhance the efficiency of DR diagnosis using fundus images. Further study with multiple non-ophthalmologist graders is needed to verify the results and strategies to improve agreement for DMO diagnosis are needed.

Choroidal structural analysis in eyes with diabetic retinopathy and diabetic macular edema-A novel OCT based imaging biomarker.

PURPOSE: To evaluate structural changes in the choroid among patients with diabetic macular edema (DME), with varying grades of diabetic retinopathy (DR), using enhance depth imaging spectral domain optical coherence tomography (EDI SD-OCT) scans. METHODS: A cross-sectional study was conducted on 82 eyes with DR and DME and 86 healthy control eyes. Eyes with DME were classified according to the severity of DR as per the international DR severity scale. Sub foveal choroidal thickness (SFCT)was obtained using EDI SD-OCT scans. These scans were binarized into luminal and stromal areas, to derive the choroidal vascularity index (CVI). CVI and SFCT were analyzed between the study and control group using paired-T test. Tukey's test was used to correlate the differences in CVI and SFCT between different grades of DR. Further analysis was done to look for the effect of DR severity and type of DME on CVI as well as SFCT using correlation coefficient and linear regression analysis. RESULTS: SFCT was significantly increased in eyes with DME as compared to the controls (334.47±51.81µm vs 284.53±56.45µm, p<0.001), and showed an ascending trend with worsening of DR, though this difference was not statistically significant [mild non-proliferative diabetic retinopathy (NPDR) = 304.33±40.39µm, moderate NPDR = 327.81±47.39µm, severe NPDR = 357.72±62.65µm, proliferative DR (PDR) = 334.59±47.4µm, p-0.09]. CVI was significantly decreased in DME with DR eyes as compared to controls (63.89±1.89 vs 67.51±2.86, p<0.001). CVI was also significantly decreased with worsening DR (mild NPDR = 66.38±0.3, moderate NPDR = 65.28±0.37, severe NPDR = 63.50±0.47, PDR = 61.27±0.9, p<0.001). CONCLUSION: SFCT and CVI are dynamic parameters that are affected by DME. Unlike CVI, SFCT is also affected by ocular and systemic factors like edema and hypertension. CVI may be a more accurate surrogate marker for DME and DR and can potentially be used to monitor the progression of DR.

Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs.

Importance: Deep learning is a family of computational methods that allow an algorithm to program itself by learning from a large set of examples that demonstrate the desired behavior, removing the need to specify rules explicitly. Application of these methods to medical imaging requires further assessment and validation. Objective: To apply deep learning to create an algorithm for automated detection of diabetic retinopathy and diabetic macular edema in retinal fundus photographs. Design and Setting: A specific type of neural network optimized for image classification called a deep convolutional neural network was trained using a retrospective development data set of 128 175 retinal images, which were graded 3 to 7 times for diabetic retinopathy, diabetic macular edema, and image gradability by a panel of 54 US licensed ophthalmologists and ophthalmology senior residents between May and December 2015. The resultant algorithm was validated in January and February 2016 using 2 separate data sets, both graded by at least 7 US board-certified ophthalmologists with high intragrader consistency. Exposure: Deep learning-trained algorithm. Main Outcomes and Measures: The sensitivity and specificity of the algorithm for detecting referable diabetic retinopathy (RDR), defined as moderate and worse diabetic retinopathy, referable diabetic macular edema, or both, were generated based on the reference standard of the majority decision of the ophthalmologist panel. The algorithm was evaluated at 2 operating points selected from the development set, one selected for high specificity and another for high sensitivity. Results: The EyePACS-1 data set consisted of 9963 images from 4997 patients (mean age, 54.4 years; 62.2% women; prevalence of RDR, 683/8878 fully gradable images [7.8%]); the Messidor-2 data set had 1748 images from 874 patients (mean age, 57.6 years; 42.6% women; prevalence of RDR, 254/1745 fully gradable images [14.6%]). For detecting RDR, the algorithm had an area under the receiver operating curve of 0.991 (95% CI, 0.988-0.993) for EyePACS-1 and 0.990 (95% CI, 0.986-0.995) for Messidor-2. Using the first operating cut point with high specificity, for EyePACS-1, the sensitivity was 90.3% (95% CI, 87.5%-92.7%) and the specificity was 98.1% (95% CI, 97.8%-98.5%). For Messidor-2, the sensitivity was 87.0% (95% CI, 81.1%-91.0%) and the specificity was 98.5% (95% CI, 97.7%-99.1%). Using a second operating point with high sensitivity in the development set, for EyePACS-1 the sensitivity was 97.5% and specificity was 93.4% and for Messidor-2 the sensitivity was 96.1% and specificity was 93.9%. Conclusions and Relevance: In this evaluation of retinal fundus photographs from adults with diabetes, an algorithm based on deep machine learning had high sensitivity and specificity for detecting referable diabetic retinopathy. Further research is necessary to determine the feasibility of applying this algorithm in the clinical setting and to determine whether use of the algorithm could lead to improved care and outcomes compared with current ophthalmologic assessment.

Defining a Minimum Set of Standardized Patient-centered Outcome Measures for Macular Degeneration.

PURPOSE: To define a minimum set of outcome measures for tracking, comparing, and improving macular degeneration care. DESIGN: Recommendations from a working group of international experts in macular degeneration outcomes registry development and patient advocates, facilitated by the International Consortium for Health Outcomes Measurement (ICHOM). METHODS: Modified Delphi technique, supported by structured teleconferences, followed by online surveys to drive consensus decisions. Potential outcomes were identified through literature review of outcomes collected in existing registries and reported in major clinical trials. Outcomes were refined by the working group and selected based on impact on patients, relationship to good clinical care, and feasibility of measurement in routine clinical practice. RESULTS: Standardized measurement of the following outcomes is recommended: visual functioning and quality of life (distance visual acuity, mobility and independence, emotional well-being, reading and accessing information); number of treatments; complications of treatment; and disease control. Proposed data collection sources include administrative data, clinical data during routine clinical visits, and patient-reported sources annually. Recording the following clinical characteristics is recommended to enable risk adjustment: age; sex; ethnicity; smoking status; baseline visual acuity in both eyes; type of macular degeneration; presence of geographic atrophy, subretinal fibrosis, or pigment epithelial detachment; previous macular degeneration treatment; ocular comorbidities. CONCLUSIONS: The recommended minimum outcomes and pragmatic reporting standards should enable standardized, meaningful assessments and comparisons of macular degeneration treatment outcomes. Adoption could accelerate global improvements in standardized data gathering and reporting of patient-centered outcomes. This can facilitate informed decisions by patients and health care providers, plus allow long-term monitoring of aggregate data, ultimately improving understanding of disease progression and treatment responses.

Proinflammatory cytokines and angiogenic and anti-angiogenic factors in vitreous of patients with proliferative diabetic retinopathy and eales' disease.

PURPOSE: To investigate the mechanism of angiogenesis in proliferative diabetic retinopathy (PDR) and Eales' disease (ED) on the basis of the levels of proinflammatory cytokines, angiogenic growth factor, and antiangiogenic factor in the vitreous humor. METHODS: Twenty-five patients with PDR, 10 patients with ED, and 25 with macular hole (MH) as control subjects were studied. The concentration of the proinflammatory cytokines interleukin-6 (IL-6), IL-8, IL-1 beta; chemokine-monocyte chemoattractant protein-1 (MCP-1); angiogenic factor-vascular endothelial growth factor (VEGF); and antiangiogenic factor-pigment epithelium derived factor (PEDF) in the vitreous fluid obtained from the eyes during vitrectomy were measured by sandwich enzyme linked immunosorbent assay (ELISA). RESULTS: IL-6, IL-8, MCP-1, and VEGF levels in the vitreous were significantly higher in PDR (P < 0.0001) and ED (P < 0.0001) than in MH patients. Conversely, the vitreous level of PEDF was significantly reduced in PDR (P < 0.0001) but not in ED. A significant correlation was observed between VEGF and IL-6 in ED patients. CONCLUSION: The authors demonstrate the importance of VEGF in retinal neovascularization of ED which is an idiopathic inflammatory venous occlusion. Further study is required to understand the interrelationship between VEGF and inflammatory cytokines in PDR and ED.

Vitrectomy for epimacular membrane secondary to adult-onset Coats' disease.

The authors describe the management of an epimacular membrane secondary to adult-onset Coats' disease. A 26-year-old man presented with decreased vision (20/120) in the right eye for 4 months. Fundus examination revealed features suggestive of Coats' disease, with a thick epimacular membrane. Optical coherence tomography revealed underlying macular thickening. The patient underwent vitrectomy with removal of the epimacular membrane and internal limiting membrane; peripheral telangiectasia were simultaneously photocoagulated. Postoperatively, his visual acuity improved to 20/20, which remained stable for 12 months. Vitrectomy yields an excellent anatomical and functional outcome in epimacular membrane due to adult-onset Coats' disease, if performed before macular exudation leads to subretinal fibrosis.

Introduction, mechanism of action and rationale for anti-vascular endothelial growth factor drugs in age-related macular degeneration.

Recent developments may provide an opportunity to improve outcome in individuals who develop neovascular age-related macular degeneration (ARMD). Several therapies have been introduced that show promise for halting the progression of this disorder. However, data from controlled clinical trials to test the relative efficacy of different management strategies across the subtypes of disease remain limited. New treatment modalities that target the neovascularization process, including leakage from choroidal neovascularization (CNV), are currently being developed. Vascular endothelial growth factor (VEGF) has been implicated as a key mediator in the pathogenesis of ARMD-related CNV. Anti-VEGF strategies show promise as potential therapeutic agents for the treatment of CNV and are currently undergoing active clinical investigation. Such strategies include anti-VEGF antibodies, anti-VEGF aptamer, gene therapy and protein kinase C inhibition. This article reviews the mechanism of action and rationale for anti-VEGF drugs in ARMD.

Combined central retinal artery and vein occlusion secondary to systemic non-Hodgkin's lymphoma.

We report a rare case of low-grade systemic B-cell non-Hodgkin's lymphoma (NHL) causing central retinal artery and vein occlusion, which was the only manifestation of disease recurrence. A young man with resolved systemic NHL underwent fluorescein angiography, magnetic resonance imaging and computed tomography to investigate a severe unilateral visual loss. A combined vascular occlusion was observed in the right eye. Neuroimaging detected optic nerve infiltration; but no systemic/ central nervous system involvement was observed. The patient was treated with high-doses of corticosteroids and optic nerve irradiation. The optic neuropathy and vascular occlusion were resistant to treatment. The subsequent neovascular glaucoma was treated by panretinal photocoagulation, which relieved the pain, but vision was not recovered. No further recurrence was observed over the following year.

Giant nodular posterior scleritis simulating choroidal melanoma.

A 45-year-old woman, complaining of sudden visual loss in the right eye (best-corrected visual acuity: 17/200), was suspected of having a malignant melanoma of the choroid. Fundus examination revealed a massive pale-colored subretinal mass temporal to the fovea, with adjoining choroidal folds and exudative retinal detachment. Ultrasonography was suggestive of posterior scleritis. Systemic evaluation was unremarkable. The patient was treated with high-dose systemic corticosteroids. The mass lesion resolved completely, with visual recovery to 20/20. Posterior scleritis should be suspected when evaluating subretinal mass-lesions, even when large and without overt inflammation. Early treatment may improve the visual prognosis and avoid misdirected management, including enucleation.

Intravitreal injection of triamcinolone acetonide for diabetic macular edema: principles and practice.

Diabetic retinopathy is fast emerging as a leading cause of newly diagnosed legal blindness amongst the working population. Macular edema, as it is commoner, accounts for more vision impairment than neovascular proliferation in diabetic patients. Laser photocoagulation, which is the standard treatment of macular edema, is associated with significant complications and an improvement in visual acuity is unsatisfactory. Intravitreal injection of corticosteroids (especially triamcinolone acetonide) is an emerging treatment modality in the management of diabetic macular edema. This article presents an overview of the principles, technique and complications associated with this procedure.

Association of VEGF and eNOS gene polymorphisms in type 2 diabetic retinopathy.

PURPOSE: Vascular endothelial growth factor (VEGF) is a major mediator of angiogenesis, and nitric oxide (NO) is an upstream and downstream regulator of VEGF mediated angiogenesis. VEGF and NO have been suggested to play an important role in the pathogenesis of microvascular complications in diabetic retinopathy (DR). The objective of this study was to examine the genetic variations of the VEGF and eNOS gene and assess their possible relationship to DR in type 2 diabetic patients in the Indian population. METHODS: In this study, 210 unrelated patients were enrolled and categorized into two study groups: a DR group, consisting of patients with proliferative diabetic retinopathy, and a diabetic without retinopathy (DWR) group comprised of patients with type 2 diabetes of more than 15 years duration who showed no signs of DR or had fewer than five dots or blot hemorrhages. Association of the genetic polymorphisms in the promoter and 5' UTR region of VEGF and the intron4 region of eNOS were studied. Total genomic DNA was isolated from peripheral blood leukocytes. PCR-RFLP analysis was performed for all samples to evaluate the genotypes. The distributions of the genotypes were compared using the chi2 test. Haplotype estimation and multiple logistic regression analysis were carried out to analyze the significance of polymorphisms. RESULTS: We investigated four reported polymorphisms in the VEGF (5' UTR, promoter) and one reported polymorphism (intron 4) in the eNOS gene in Type 2 diabetes patients with (n=120) and without (n=90) retinopathy. The genotype distribution of the C(-7)T, T(-1498)C, and C(-634)G polymorphisms of VEGF differed significantly between patients with DR and DWR (p=0.001, p=0.0001, and p=0.021, respectively). Allele C in the -1498 region (p=0.0001) and T in -7 region (p=0.002) were also found to be significantly increased in patients with retinopathy. Calculated odds ratios (OR) for three heterozygous genotypes of C(-7)T, T(-1498)C, and C(-634)G regions were 4.17 (95% CI: 1.90-9.18, p=0.0001), 4.37 (95% CI: 2.44-7.84, p=0.0001), and 2.33 (95% CI: 1.24-4.36, p=0.008), respectively, and was found to be significantly higher in the DR group when compared with the DWR group. Multiple logistic regression analysis revealed that the nongenetic parameters, age (p=0.024) and duration of diabetes (p=0.009), and the genetic parameters, like VEGF C(-7)T (p=0.002) and T(-1498)C (p=0.001) polymorphisms, were significantly associated with DR. The frequencies of haplotype consisting of the majority of alleles in VEGF were found to be significantly associated with DR. The genotype distribution of eNOS did not differ significantly between the two study groups, and therefore the eNOS intron 4 polymorphism was considered to be less significant. CONCLUSIONS: This is the first study to report VEGF and eNOS gene polymorphisms in patients with DR in the Indian population. The data suggest that the polymorphisms in the 5' UTR and promoter region of VEGF could be regarded as a major genetic risk factor for DR.

Macular hole formation and spontaneous closure after vitrectomy for central retinal vein occlusion.

PURPOSE: The purpose was to describe the formation and spontaneous closure of an iatrogenic macular hole after vitrectomy for central retinal vein occlusion (CRVO). METHODS: A 45-year-old woman presented with loss of vision in the right eye due to CRVO. The left eye was normal. Best-corrected visual acuity (BCVA) was 3/60 in the right eye and 6/6 in the left. She underwent vitrectomy and radial optic neurotomy (RON) in the right eye. Fluorescein angiography and optical coherence tomography (OCT) were performed pre-operatively, on the first post-operative day, and subsequently at 1, 3, 6 and 12 months. RESULTS: There were no intra-operative complications. On the first post-operative day, fundus examination of the right eye revealed a small macular hole, confirmed by OCT. No additional intervention was carried out. Examination after a month revealed clearing of retinal haemorrhages, resolution of macular oedema and spontaneous closure of the macular hole. The closure was maintained until the last follow-up a year later. BCVA improved to 6/24 by 3 months, and was stable over the follow-up period. The left eye remained unchanged. CONCLUSION: While performing vitrectomy for CRVO, the induction of posterior vitreous detachment may deroof a foveal cyst in the highly oedematous macula to form a macular hole. A small macular hole, however, may close spontaneously without further intervention.

Nocardia endophthalmitis after cataract surgery: clinicomicrobiological study.

PURPOSE: To analyze the clinical presentation, microbiological profile, and treatment outcome in cases of nocardial endophthalmitis after cataract surgery. DESIGN: Retrospective, observational case series. METHODS: setting: Tertiary referral hospital. patients: Retrospective analysis of 24 cases of culture-proven exogenous nocardial endophthalmitis following cataract surgery between January 2000 and December 2003. main outcome measures: Visual acuity and anatomic outcome. RESULTS: Visual outcome was poor in most patients, with 79% (19/24) of patients obtaining a final visual acuity of hand motions or worse and 46% (11/24) of patients with no light perception. Initial presenting visual acuity was strongly associated with final visual outcome (P = .0026). There was a statistically significant trend favoring better visual outcome in patients who presented early after the onset of symptoms (P = .01). The majority of the eyes (23/24) had an early onset with predominantly anterior segment involvement. Wound infection was noted in 46% (11/24) of the patients at presentation. Clinical features included scleral abscess, cotton ball exudates in the anterior chamber, fluffy exudates on the corneal endothelium, nodular exudates on the iris, and hypopyon. Among the specimens cultured, the highest positivity was seen with the anterior chamber aspirate (15/16 = 93.75%), which was significantly higher than the vitreous aspirate cultured (1/22 = 4.54%) (P = .001). CONCLUSION: Nocardia endophthalmitis after cataract surgery is an aggressive disease with poor visual prognosis. Early diagnosis and treatment with good visual acuity at presentation may be associated with better visual outcome.

Optical coherence tomography in presumed congenital simple hamartoma of retinal pigment epithelium.

PURPOSE: To report the optical coherence tomography (OCT) findings in two cases of presumed congenital simple retinal pigment epithelial (RPE) hamartoma. DESIGN: Observational case report. METHODS: Fundus examination, fluorescein angiography, ultrasonography, and OCT were done on two cases of simple RPE hamartoma. RESULTS: A heavily pigmented solitary macular lesion was noted in the right eye of a 22-year-old man and a 55-year-old woman. The young man had a foveal lesion and a vision of 10/200; the woman with eccentric lesion had a vision of 20/20. The lesions blocked fluorescence on angiogram; sonography showed hyper-reflective nodules. OCT demonstrated full-thickness retinal replacement, complete optical shadowing, and abrupt margins of the lesion in both the cases, and vitreomacular adhesion at the temporal margin of the foveal lesion. CONCLUSIONS: Congenital simple RPE hamartoma may rarely occur at the foveal center, resulting in poor visual acuity. OCT is a useful non-invasive adjunct for diagnosis of this rare tumor and may reveal additional features like vitreoretinal adhesion.

A pilot study on awareness of diabetic retinopathy among non-medical persons in South India. The challenge for eye care programmes in the region.

This study was conducted to determine awareness and practices relating to diabetic retinopathy among non-medical persons in a south Indian population. In this population-based cross-sectional study, trained social workers conducted face-to-face interviews using a semi-structured questionnaire with 200 randomly selected paramedical personnel and 204 persons randomly selected from the community. Responses were graded on a five-point scale. Over half of respondents were not aware of risk factors for diabetic retinopathy. Only one-fifth of paramedics and one-tenth of persons from the community were aware that uncontrolled diabetes was a risk factor for retinopathy. Over 75% of respondents were not aware of either laser or surgery as an intervention for retinopathy. Although 80% of respondents from the community felt that yearly eye examinations were essential, only 43.5% had ever visited an ophthalmologist. Nearly three-fourths of paramedical personnel did not have any material related to diabetes for health education. It is evident that considerable effort is required to improve awareness of diabetic retinopathy, and to translate this improved awareness to actual utilisation of services.