Clinically significant macular edema in an underserved population: Association with demographic factors and hemoglobin A1c.

SIGNIFICANCE: Suspected clinically significant macular edema (SCSME) from exudates differed among ethnic groups in our underserved population. African American and Asian subjects had higher prevalence than Hispanics and non-Hispanic Caucasians, from the same clinics. Men had higher prevalence than women. Highly elevated blood glucose was frequent and associated with SCSME. PURPOSE: We investigated the association between the presence of SCSME from exudates and hemoglobin A1c (HbA1c), as well as demographic factors such as age, sex, and ethnic group. Our population was underserved diabetic patients from the same geographic locations. Ethnic groups were White Hispanic, non-Hispanic Caucasian, African American, and Asian, with a high proportion of underrepresented minorities. METHODS: In a diabetic retinopathy screening study at four community clinics in Alameda County, California, nonmydriatic 45° color fundus images were collected from underserved diabetic subjects following the EyePACS imaging protocol. Images were analyzed for SCSME from exudates by two certified graders. Logistic regression assessed the association between SCSME from exudates and age, sex, ethnic group, and HbA1c. RESULTS: Of 1997 subjects, 147 (7.36%) had SCSME from exudates. The mean ± standard deviation age was 53.4 ± 10.5 years. The mean ± standard deviation HbA1c level was 8.26 ± 2.04. Logistic regression analysis indicated a significant association between presence of SCSME from exudates and HbA1c levels (p<0.001), sex (p=0.027), and ethnicity (p=0.030). African Americans (odds ratio [OR], 1.63; 95% confidence interval [CI], 1.06 to 2.50; p=0.025) and Asians (OR, 1.63; 95% CI, 1.05 to 2.54; p=0.029) had a higher risk than Hispanics. After adjusting for ethnicity, sex, and age, the odds of developing SCSME from exudates increased by 26.5% with every 1% increase in HbA1c level (OR, 1.26; 95% CI, 1.18 to 1.36; p<0.001). CONCLUSIONS: In our underserved population, many diabetic patients had very high HbA1c values. Ethnic background (African American > Asians > Hispanics), sex (male > female), and HbA1c level were strong indicators for identifying who is at increased risk of developing SCSME from exudates.

Diabetic Retinopathy during pregnancy in Hispanic women with latent Toxoplasma gondii infection.

BACKGROUND: Retinal photography was performed in pregnancy and postpartum in pregnant Hispanic women with latent Toxoplasma gondii (TG) infection in order to screen for characteristic retinal lesions or the particular scars found in people with active T. gondii infection. A comparison group of TG negative women was included in the study but they did not have retinal photography. OBJECTIVE: The goal of the parent study was to assess for adverse pregnancy events and evidence for parasite reactivation in TG positive (TG + ) women, through examination of the eyes for characteristic lesions. Retinal photography, usually at prenatal visits 2 (17 +/- 3.35 weeks) and 3 (26.3+/-1.75) weeks, was done on TG + women. Fifty-six of these women also (43 %) had retinal photography at the postpartum visit. Health and demographic data were obtained at the first prenatal visit for all women. STUDY DESIGN: From the 690 recruited at the first prenatal visit, 128 TG- women and 158 TG + women were enrolled in a prospective study through pregnancy and the postpartum. All TG- women (n = 532) provided data at the first prenatal visit and throughout their pregnancy and birth through the EHR. This allowed comparison of health and outcome data for the TG + compared to a larger number of TG- Hispanic pregnant women. RESULTS: While there was no evidence of ocular toxoplasmosis during pregnancy, there was a surprisingly large number (42 %) of TG + women with diabetic retinopathy (DR). We also observed that TG + women had a 20 % incidence of gestational diabetes mellitus (GDM) compared to 11.3 % in the TG- women (p = 0.01). At postpartum (mean 5.6 weeks), 23 of 30 women with pregnancy DR showed no DR in the postpartum. CONCLUSIONS: No characteristic T. gondii lesions were discovered. Retinal photography serendipitously revealed DR in these T. gondii positive women. It was also found that latent TG infection was associated with increased incidence of GDM. Hispanic pregnant women's increased risk for latent TG infection, GDM and DR are underappreciated. Retinal photography may need to be considered an innovative approach to screening.

A deep learning model for novel systemic biomarkers in photographs of the external eye: a retrospective study.

BACKGROUND: Photographs of the external eye were recently shown to reveal signs of diabetic retinal disease and elevated glycated haemoglobin. This study aimed to test the hypothesis that external eye photographs contain information about additional systemic medical conditions. METHODS: We developed a deep learning system (DLS) that takes external eye photographs as input and predicts systemic parameters, such as those related to the liver (albumin, aspartate aminotransferase [AST]); kidney (estimated glomerular filtration rate [eGFR], urine albumin-to-creatinine ratio [ACR]); bone or mineral (calcium); thyroid (thyroid stimulating hormone); and blood (haemoglobin, white blood cells [WBC], platelets). This DLS was trained using 123 130 images from 38 398 patients with diabetes undergoing diabetic eye screening in 11 sites across Los Angeles county, CA, USA. Evaluation focused on nine prespecified systemic parameters and leveraged three validation sets (A, B, C) spanning 25 510 patients with and without diabetes undergoing eye screening in three independent sites in Los Angeles county, CA, and the greater Atlanta area, GA, USA. We compared performance against baseline models incorporating available clinicodemographic variables (eg, age, sex, race and ethnicity, years with diabetes). FINDINGS: Relative to the baseline, the DLS achieved statistically significant superior performance at detecting AST >36·0 U/L, calcium <8·6 mg/dL, eGFR <60·0 mL/min/1·73 m2, haemoglobin <11·0 g/dL, platelets <150·0 × 103/µL, ACR ≥300 mg/g, and WBC <4·0 × 103/µL on validation set A (a population resembling the development datasets), with the area under the receiver operating characteristic curve (AUC) of the DLS exceeding that of the baseline by 5·3-19·9% (absolute differences in AUC). On validation sets B and C, with substantial patient population differences compared with the development datasets, the DLS outperformed the baseline for ACR ≥300·0 mg/g and haemoglobin <11·0 g/dL by 7·3-13·2%. INTERPRETATION: We found further evidence that external eye photographs contain biomarkers spanning multiple organ systems. Such biomarkers could enable accessible and non-invasive screening of disease. Further work is needed to understand the translational implications. FUNDING: Google.

Artificial Intelligence for Predicting and Diagnosing Complications of Diabetes.

Artificial intelligence can use real-world data to create models capable of making predictions and medical diagnosis for diabetes and its complications. The aim of this commentary article is to provide a general perspective and present recent advances on how artificial intelligence can be applied to improve the prediction and diagnosis of six significant complications of diabetes including (1) gestational diabetes, (2) hypoglycemia in the hospital, (3) diabetic retinopathy, (4) diabetic foot ulcers, (5) diabetic peripheral neuropathy, and (6) diabetic nephropathy.

Detection of signs of disease in external photographs of the eyes via deep learning.

Retinal fundus photographs can be used to detect a range of retinal conditions. Here we show that deep-learning models trained instead on external photographs of the eyes can be used to detect diabetic retinopathy (DR), diabetic macular oedema and poor blood glucose control. We developed the models using eye photographs from 145,832 patients with diabetes from 301 DR screening sites and evaluated the models on four tasks and four validation datasets with a total of 48,644 patients from 198 additional screening sites. For all four tasks, the predictive performance of the deep-learning models was significantly higher than the performance of logistic regression models using self-reported demographic and medical history data, and the predictions generalized to patients with dilated pupils, to patients from a different DR screening programme and to a general eye care programme that included diabetics and non-diabetics. We also explored the use of the deep-learning models for the detection of elevated lipid levels. The utility of external eye photographs for the diagnosis and management of diseases should be further validated with images from different cameras and patient populations.

Feasibility of a Bi-national Consulate-based Teleophthalmology Screening and Referral Protocol for Diabetic Retinopathy among Mexican Migrants in the United States.

AIMS: To evaluate a bi-national consulate-based teleophthalmology screening service for diabetic retinopathy (DR) among Mexican migrants in the U.S. METHODS: Adult visitors (n=508) at Mexican consulates in California with self-reported diabetes underwent questionnaires and fundus photography. Photographs were graded for DR by retina fellows in Mexico via teleophthalmology. Participants were contacted with results and provided referrals when necessary. RESULTS: Nearly all (97.6%) participants were aware that diabetes can cause vision loss. One-quarter (24.4%) had undergone an eye examination in the past year. Barriers to care were cost (53.9%) and insurance (45.6%). Most (85.4-91.1%) reported that Spanish-speaking providers and provision of screening in primary care would increase participation in screening. Any DR, vision-threatening DR, or proliferative DR were found in 30.2%, 9.9%, and 5.4% of participants, respectively. Nearly one-fifth (19.5%) received referrals. CONCLUSIONS: Screening in Mexican consulates may improve DR detection and treatment among Mexican migrants in the U.S.

Deep Learning to Detect OCT-derived Diabetic Macular Edema from Color Retinal Photographs: A Multicenter Validation Study.

PURPOSE: To validate the generalizability of a deep learning system (DLS) that detects diabetic macular edema (DME) from 2-dimensional color fundus photographs (CFP), for which the reference standard for retinal thickness and fluid presence is derived from 3-dimensional OCT. DESIGN: Retrospective validation of a DLS across international datasets. PARTICIPANTS: Paired CFP and OCT of patients from diabetic retinopathy (DR) screening programs or retina clinics. The DLS was developed using data sets from Thailand, the United Kingdom, and the United States and validated using 3060 unique eyes from 1582 patients across screening populations in Australia, India, and Thailand. The DLS was separately validated in 698 eyes from 537 screened patients in the United Kingdom with mild DR and suspicion of DME based on CFP. METHODS: The DLS was trained using DME labels from OCT. The presence of DME was based on retinal thickening or intraretinal fluid. The DLS's performance was compared with expert grades of maculopathy and to a previous proof-of-concept version of the DLS. We further simulated the integration of the current DLS into an algorithm trained to detect DR from CFP. MAIN OUTCOME MEASURES: The superiority of specificity and noninferiority of sensitivity of the DLS for the detection of center-involving DME, using device-specific thresholds, compared with experts. RESULTS: The primary analysis in a combined data set spanning Australia, India, and Thailand showed the DLS had 80% specificity and 81% sensitivity, compared with expert graders, who had 59% specificity and 70% sensitivity. Relative to human experts, the DLS had significantly higher specificity (P = 0.008) and noninferior sensitivity (P < 0.001). In the data set from the United Kingdom, the DLS had a specificity of 80% (P < 0.001 for specificity of >50%) and a sensitivity of 100% (P = 0.02 for sensitivity of > 90%). CONCLUSIONS: The DLS can generalize to multiple international populations with an accuracy exceeding that of experts. The clinical value of this DLS to reduce false-positive referrals, thus decreasing the burden on specialist eye care, warrants a prospective evaluation.

Predicting the risk of developing diabetic retinopathy using deep learning.

BACKGROUND: Diabetic retinopathy screening is instrumental to preventing blindness, but scaling up screening is challenging because of the increasing number of patients with all forms of diabetes. We aimed to create a deep-learning system to predict the risk of patients with diabetes developing diabetic retinopathy within 2 years. METHODS: We created and validated two versions of a deep-learning system to predict the development of diabetic retinopathy in patients with diabetes who had had teleretinal diabetic retinopathy screening in a primary care setting. The input for the two versions was either a set of three-field or one-field colour fundus photographs. Of the 575 431 eyes in the development set 28 899 had known outcomes, with the remaining 546 532 eyes used to augment the training process via multitask learning. Validation was done on one eye (selected at random) per patient from two datasets: an internal validation (from EyePACS, a teleretinal screening service in the USA) set of 3678 eyes with known outcomes and an external validation (from Thailand) set of 2345 eyes with known outcomes. FINDINGS: The three-field deep-learning system had an area under the receiver operating characteristic curve (AUC) of 0·79 (95% CI 0·77-0·81) in the internal validation set. Assessment of the external validation set-which contained only one-field colour fundus photographs-with the one-field deep-learning system gave an AUC of 0·70 (0·67-0·74). In the internal validation set, the AUC of available risk factors was 0·72 (0·68-0·76), which improved to 0·81 (0·77-0·84) after combining the deep-learning system with these risk factors (p<0·0001). In the external validation set, the corresponding AUC improved from 0·62 (0·58-0·66) to 0·71 (0·68-0·75; p<0·0001) following the addition of the deep-learning system to available risk factors. INTERPRETATION: The deep-learning systems predicted diabetic retinopathy development using colour fundus photographs, and the systems were independent of and more informative than available risk factors. Such a risk stratification tool might help to optimise screening intervals to reduce costs while improving vision-related outcomes. FUNDING: Google.

The Real-World Impact of Artificial Intelligence on Diabetic Retinopathy Screening in Primary Care.

The study by Shah et al published in this issue of the Journal of Diabetes Science and Technology validates the IDx autonomous diabetic retinopathy (DR) screening program in a real-world setting. The study found high sensitivity (100%) but low specificity (82%) for referable DR. The resulting positive predictive value of 19% means that four out of five patients without referable DR would be referred to ophthalmology causing a significant burden to ophthalmologists, primary care clinics, and patients. Artificial intelligence programs that provide better specificity, multiple levels of DR, and annotations of where lesions are located in the retina may function better than a simple referral/no referral output. This will allow for better engagement of patients through the difficult process of adhering to treatment recommendations and control their diabetes.

Adherence to ophthalmology referral, treatment and follow-up after diabetic retinopathy screening in the primary care setting.

INTRODUCTION: Telemedicine-based diabetic retinopathy screening (DRS) in primary care settings has increased the screening rates of patients with diabetes. However, blindness from vision-threatening diabetic retinopathy (VTDR) is a persistent problem. This study examined the extent of patients' adherence to postscreening recommendations. RESEARCH DESIGN/METHODS: A retrospective record review was conducted in primary care clinics of a large county hospital in the USA. All patients with diabetes detected with VTDR in two time periods, differing in record type used, were included in the study: 2012-2014, paper charts only; 2015-2017, combined paper charts/electronic medical records (EMRs), or EMRs only. Adherence rates for keeping initial ophthalmology appointments, starting recommended treatments, and keeping follow-up appointments were determined. RESULTS: Adequate records were available for 6046 patients; 408 (7%) were detected with VTDR and recommended for referral to ophthalmology. Only 5% completed a first ophthalmology appointment within recommended referral interval, 15% within twice the recommended interval, and 51% within 1 year of DRS. Patients screened in 2015-2017 were more likely to complete a first ophthalmology appointment than those in 2012-2014. Ophthalmic treatment was recommended in half of the patients, of whom 94% initiated treatment. A smaller percentage (41%) adhered completely to post-treatment follow-up. Overall, 28% of referred patients: (1) kept a first ophthalmology appointment; (2) were recommended for treatment; and (3) initiated the treatment. Most patients failing to keep first ophthalmology appointments continued non-ophthalmic medical care at the institution. EMRs provided more complete information than paper charts. CONCLUSIONS: Reducing vision impairment from VTDR requires greater emphasis on timely adherence to ophthalmology referral and follow-up. Prevention of visual loss from VTDR starts with retinopathy screening, but must include patient engagement, adherence monitoring, and streamlining ophthalmic referral and management. Revision of these processes has already been implemented at the study site, incorporating lessons from this investigation.

Predicting optical coherence tomography-derived diabetic macular edema grades from fundus photographs using deep learning.

Center-involved diabetic macular edema (ci-DME) is a major cause of vision loss. Although the gold standard for diagnosis involves 3D imaging, 2D imaging by fundus photography is usually used in screening settings, resulting in high false-positive and false-negative calls. To address this, we train a deep learning model to predict ci-DME from fundus photographs, with an ROC-AUC of 0.89 (95% CI: 0.87-0.91), corresponding to 85% sensitivity at 80% specificity. In comparison, retinal specialists have similar sensitivities (82-85%), but only half the specificity (45-50%, p < 0.001). Our model can also detect the presence of intraretinal fluid (AUC: 0.81; 95% CI: 0.81-0.86) and subretinal fluid (AUC 0.88; 95% CI: 0.85-0.91). Using deep learning to make predictions via simple 2D images without sophisticated 3D-imaging equipment and with better than specialist performance, has broad relevance to many other applications in medical imaging.

Diabetic Retinopathy and the Cascade into Vision Loss.

Vision loss from diabetic retinopathy should be unnecessary for patients with access to diabetic retinopathy screening, yet it still occurs at high rates and in varied contexts. Precisely because vision loss is only one of many late-stage complications of diabetes, interfering with the management of diabetes and making self-care more difficult, Vision Threatening Diabetic Retinopathy (VTDR) is considered a "high stakes" diagnosis. Our mixed-methods research addressed the contexts of care and treatment seeking in a sample of people with VTDR using safety-net clinic services and eye specialist referrals. We point to conceptual weaknesses in the single disease framework of health care by diagnosis, and we use the framework of "cascades" to clarify why and how certain non-clinical factors come to bear on long-term experiences of complex chronic diseases.

The Value of Automated Diabetic Retinopathy Screening with the EyeArt System: A Study of More Than 100,000 Consecutive Encounters from People with Diabetes.

Background: Current manual diabetic retinopathy (DR) screening using eye care experts cannot scale to screen the growing population of diabetes patients who are at risk for vision loss. EyeArt system is an automated, cloud-based artificial intelligence (AI) eye screening technology designed to easily detect referral-warranted DR immediately through automated analysis of patient's retinal images. Methods: This retrospective study assessed the diagnostic efficacy of the EyeArt system v2.0 analyzing 850,908 fundus images from 101,710 consecutive patient visits, collected from 404 primary care clinics. Presence or absence of referral-warranted DR (more than mild nonproliferative DR [NPDR]) was automatically detected by the EyeArt system for each patient encounter, and its performance was compared against a clinical reference standard of quality-assured grading by rigorously trained certified ophthalmologists and optometrists. Results: Of the 101,710 visits, 75.7% were nonreferable, 19.3% were referable to an eye care specialist, and in 5.0%, the DR level was unknown as per the clinical reference standard. EyeArt screening had 91.3% (95% confidence interval [CI]: 90.9-91.7) sensitivity and 91.1% (95% CI: 90.9-91.3) specificity. For 5446 encounters with potentially treatable DR (more than moderate NPDR and/or diabetic macular edema), the system provided a positive "refer" output to 5363 encounters achieving sensitivity of 98.5%. Conclusions: This study captures variations in real-world clinical practice and shows that an AI DR screening system can be safe and effective in the real world. This study demonstrates the value of this easy-to-use, automated tool for endocrinologists, diabetologists, and general practitioners to address the growing need for DR screening and monitoring.

Central Macular Thickness in Diabetic Patients: A Sex-based Analysis.

SIGNIFICANCE: The pathological changes in clinically significant diabetic macular edema lead to greater retinal thickening in men than in women. Therefore, male sex should be considered a potential risk factor for identifying individuals with the most severe pathological changes. Understanding this excessive retinal thickening in men may help preserve vision. PURPOSE: The purpose of this study was to investigate the sex differences in retinal thickness in diabetic patients. We tested whether men with clinically significant macular edema had even greater central macular thickness than expected from sex differences without significant pathological changes. This study also aimed to determine which retinal layers contribute to abnormal retinal thickness. METHODS: From 2047 underserved adult diabetic patients from Alameda County, CA, 142 patients with clinically significant macular edema were identified by EyePACS-certified graders using color fundus images (Canon CR6-45NM). First, central macular thickness from spectral domain optical coherence tomography (iVue; Optovue Inc.) was compared in 21 men versus 21 women without clinically significant macular edema. Then, a planned comparison contrasted the greater values of central macular thickness in men versus women with clinically significant macular edema as compared with those without. Mean retinal thickness and variability of central macular layers were compared in men versus women. RESULTS: Men without clinically significant macular edema had a 12-µm greater central macular thickness than did women (245 ± 21.3 and 233 ± 13.4 µm, respectively; t40 = -2.18, P = .04). Men with clinically significant macular edema had a 67-µm greater central macular thickness than did women (383 ± 48.7 and 316 ± 60.4 µm, P < .001); that is, men had 55 µm or more than five times more (t20 = 2.35, P = .02). In men, the outer-nuclear-layer thickness was more variable, F10,10 = 9.34. CONCLUSIONS: Underserved diabetic men had thicker retinas than did women, exacerbated by clinically significant macular edema.

Blind spots in telemedicine: a qualitative study of staff workarounds to resolve gaps in diabetes management.

BACKGROUND: Novel telemedicine platforms have expanded access to critical retinal screening into primary care settings. This increased access has contributed to improved retinal screening uptake for diabetic patients, particularly those treated in Federally Qualified Health Centers ('safety net' clinics). The aim of this study was to understand how the implementation of telemedical screening for diabetic retinopathy within primary care settings is improving the delivery of critical preventative services, while also introducing changes into clinic workflows and creating additional tasks and responsibilities within resource-constrained clinics. METHODS: A qualitative approach was employed to track workflows and perspectives from a range of medical personnel involved in the telemedicine platform for diabetic retinopathy screening and subsequent follow-up treatment. Data were collected through semi-structured interviews and participant observation at three geographically-dispersed Federally Qualified Health Centers in California. Qualitative analysis was performed using standard thematic analytic approaches within a qualitative data analysis software program. RESULTS: The introduction of telemedicine platforms, such as diabetic retinopathy screening, into primary care settings is creating additional strain on medical personnel across the diabetes eye care management spectrum. Central issues are related to scheduling patients, issuing referrals for follow-up care and treatment, and challenges to improving adherence to treatment and diabetes management. These issues are overcome in many cases through workarounds, or when medical staff work outside of their job descriptions, purview, and permission to move patients through the diabetes management continuum. CONCLUSIONS: This study demonstrates how the implementation of a novel telemedical platform for diabetic retinopathy screening contributes to the phenomenon of workarounds that account for additional tasks and patient volume. These workarounds should not be considered a sustainable model of health care delivery, but rather as an initial step to understanding where issues are and how clinics can adapt to the inclusion of telemedicine and ultimately increase access to care. The presence of workarounds suggests that as telemedicine is expanded, adequate resources, as well as collaborative, cross-sectoral co-design of new workflows must be simultaneously provided. Systematic bolstering of resources would contribute to more consistent success of telemedicine screening platforms and improved treatment and prevention of disease-related complications.

Automated Quality Assessment of Colour Fundus Images for Diabetic Retinopathy Screening in Telemedicine.

Fundus images obtained in a telemedicine program are acquired at different sites that are captured by people who have varying levels of experience. These result in a relatively high percentage of images which are later marked as unreadable by graders. Unreadable images require a recapture which is time and cost intensive. An automated method that determines the image quality during acquisition is an effective alternative. To determine the image quality during acquisition, we describe here an automated method for the assessment of image quality in the context of diabetic retinopathy. The method explicitly applies machine learning techniques to access the image and to determine 'accept' and 'reject' categories. 'Reject' category image requires a recapture. A deep convolution neural network is trained to grade the images automatically. A large representative set of 7000 colour fundus images was used for the experiment which was obtained from the EyePACS that were made available by the California Healthcare Foundation. Three retinal image analysis experts were employed to categorise these images into 'accept' and 'reject' classes based on the precise definition of image quality in the context of DR. The network was trained using 3428 images. The method shows an accuracy of 100% to successfully categorise 'accept' and 'reject' images, which is about 2% higher than the traditional machine learning method. On a clinical trial, the proposed method shows 97% agreement with human grader. The method can be easily incorporated with the fundus image capturing system in the acquisition centre and can guide the photographer whether a recapture is necessary or not.

Can Commercially Available Handheld Retinal Cameras Effectively Screen Diabetic Retinopathy?

Organizations that care for people with diabetes have increasingly adopted telemedicine-based diabetic retinopathy screening (TMDRS) as a way to increase adherence to recommended retinal exams. Recently, handheld retinal cameras have emerged as a low-cost, lightweight alternative to traditional bulky tabletop retinal cameras. Few published clinical trials have been performed on handheld retinal cameras. Peer-reviewed articles about commercially available handheld retinal cameras have concluded that they are a usable alternative for TMDRS, however, the clinical results presented in these articles do not meet criteria published by the United Kingdom Diabetic Eye Screening Programme and the American Academy of Ophthalmology. The future will likely remedy the shortcomings of currently available handheld retinal cameras, and will create more opportunities for preventing diabetic blindness.

A Revised Approach for the Detection of Sight-Threatening Diabetic Macular Edema.

IMPORTANCE: Diabetic macular edema is one of the leading causes of vision loss among working-age adults in the United States. Telemedicine screening programs and epidemiological studies rely on monoscopic fundus photography for the detection of clinically significant macular edema (CSME). Improving the accuracy of detecting CSME from monoscopic images could be valuable while recognizing the limitations of such detection in an era of optical coherence tomography detection of diabetic macular edema. OBJECTIVE: To evaluate the screening test accuracy of radially arranged sectors affected by hard exudates in the detection of CSME. DESIGN, SETTING, AND PARTICIPANTS: This investigation was a cross-sectional study of CSME grading in monoscopic images using a sectors approach. The Early Treatment Diabetic Retinopathy Study criteria were used to confirm the presence of CSME by the following 2 methods: stereoscopic fundus photography (method 1) and dilated biomicroscopy in combination with optical coherence tomography (method 2). Participants were recruited at a university-based practice between June 14, 2014, and December 28, 2015. MAIN OUTCOMES AND MEASURES: Area under the receiver operating characteristic curve, sensitivity, specificity, positive predictive value, and negative predictive value. RESULTS: A total of 207 eyes from an ethnically/racially diverse group of 207 patients (mean [SD] age, 53.6 [10.8] years; 58.9% [122 of 207] female) were included in the analysis. Twelve eyes (5.8%) were diagnosed as having CSME based on method 1. The intermethod and intergrader agreement for CSME diagnosis and sector count was substantial (κ range, 0.66 [95% CI, 0.47-0.85] to 0.75 [95% CI, 0.53-0.97]; P < .001 for all). Area under the receiver operating characteristic curve was 93.2% (95% CI, 84.2%-100%) when evaluating a sectors approach against method 1 as a reference test and offered up to an 8.6% (95% CI, 3.0%-14.3%) increase in specificity compared with the existing methods of detection. The positive predictive value was 33.3% (95% CI, 25.6%-45.5%), and the negative predictive value was 98.1% (95% CI, 96.9%-100%). The results were similar when comparing a sectors approach with method 2 as a reference test. CONCLUSIONS AND RELEVANCE: A sectors approach shows good screening test characteristics for the detection of CSME. Its implementation in the existing telemedicine programs would require minimal resources. This approach will have the greatest effect in a setting where implementation of optical coherence tomography, a more objective and sensitive way to detect retinal thickening, is not feasible. The proposed method also may be easily incorporated in the automated diabetic retinopathy detection algorithms.

Comparison of Cysts in Red and Green Images for Diabetic Macular Edema.

PURPOSE: To investigate whether cysts in diabetic macular edema are better visualized in the red channel of color fundus camera images, as compared with the green channel, because color fundus camera screening methods that emphasize short-wavelength light may miss cysts in patients with dark fundi or changes to outer blood retinal barrier. METHODS: Fundus images for diabetic retinopathy photoscreening were acquired for a study with Aeon Imaging, EyePACS, University of California Berkeley, and Indiana University. There were 2047 underserved, adult diabetic patients, of whom over 90% self-identified as a racial/ethnic identify other than non-Hispanic white. Color fundus images at nominally 45 degrees were acquired with a Canon Cr-DGi non-mydriatic camera (Tokyo, Japan) then graded by an EyePACS certified grader. From the 148 patients graded to have clinically significant macular edema by the presence of hard exudates in the central 1500 µm of the fovea, we evaluated macular cysts in 13 patients with cystoid macular edema. Age ranged from 33 to 68 years. Color fundus images were split into red, green, and blue channels with custom Matlab software (Mathworks, Natick, MA). The diameter of a cyst or confluent cysts was quantified in the red-channel and green-channel images separately. RESULTS: Cyst identification gave complete agreement between red-channel images and the standard full-color images. This was not the case for green-channel images, which did not expose cysts visible with standard full-color images in five cases, who had dark fundi. Cysts appeared more numerous and covered a larger area in the red channel (733 ± 604 µm) than in the green channel (349 ± 433 µm, P < .006). CONCLUSIONS: Cysts may be underdetected with the present fundus camera methods, particularly when short-wavelength light is emphasized or in patients with dark fundi. Longer wavelength techniques may improve the detection of cysts and provide more information concerning the early stages of diabetic macular edema or the outer blood retinal barrier.

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.