Surveillance of people with previously successfully treated diabetic macular oedema and proliferative diabetic retinopathy by trained ophthalmic graders: cost analysis from the EMERALD study.

BACKGROUND/AIMS: Surveillance of people with previously successfully treated diabetic macular oedema (DMO) and proliferative diabetic retinopathy (PDR) adds pressure on ophthalmology services. This study evaluated a new surveillance pathway entailing multimodal imaging reviewed by trained ophthalmic graders and compared it with the current standard care (face-to-face evaluation by an ophthalmologist). METHODS: Cost analysis of the new ophthalmic grader pathway, compared with the standard of care, from the perspective of the UK National Health Service, based on evidence from the Effectiveness of Multimodal imaging for the Evaluation of Retinal oedema And new vesseLs in Diabetic retinopathy study. Resource use data were prospectively obtained including times to undertake each procedure. Effectiveness was assessed in terms of sensitivity and specificity of referral decisions in the grader pathway. Costs (SDs) were analysed per 100 patients separately for DMO and PDR at 2018/2019 costs. RESULTS: For DMO, where sensitivity was very high (97%), the cost difference (savings) for the grader's pathway would be £1390 per 100 patients. For PDR, the cost would be reduced by £461 for seven-field Early Treatment for Diabetic Retinopathy Study (ETDRS) images and by £1889 for ultrawide field images, per 100 patients. Ultrawide images required less time to be obtained and read than seven-field ETDRS. The real savings would be in ophthalmologist time, which could be then redirected to the evaluation of people at high risk of visual loss. CONCLUSIONS: Surveillance of people with previously successfully treated DMO and PDR by trained ophthalmic graders can achieve satisfactory results and release ophthalmologist time. TRIAL REGISTRATION NUMBERS: NCT03490318, ISRCTN10856638.

Prospective evaluation of an artificial intelligence-enabled algorithm for automated diabetic retinopathy screening of 30 000 patients.

BACKGROUND/AIMS: Human grading of digital images from diabetic retinopathy (DR) screening programmes represents a significant challenge, due to the increasing prevalence of diabetes. We evaluate the performance of an automated artificial intelligence (AI) algorithm to triage retinal images from the English Diabetic Eye Screening Programme (DESP) into test-positive/technical failure versus test-negative, using human grading following a standard national protocol as the reference standard. METHODS: Retinal images from 30 405 consecutive screening episodes from three English DESPs were manually graded following a standard national protocol and by an automated process with machine learning enabled software, EyeArt v2.1. Screening performance (sensitivity, specificity) and diagnostic accuracy (95% CIs) were determined using human grades as the reference standard. RESULTS: Sensitivity (95% CIs) of EyeArt was 95.7% (94.8% to 96.5%) for referable retinopathy (human graded ungradable, referable maculopathy, moderate-to-severe non-proliferative or proliferative). This comprises sensitivities of 98.3% (97.3% to 98.9%) for mild-to-moderate non-proliferative retinopathy with referable maculopathy, 100% (98.7%,100%) for moderate-to-severe non-proliferative retinopathy and 100% (97.9%,100%) for proliferative disease. EyeArt agreed with the human grade of no retinopathy (specificity) in 68% (67% to 69%), with a specificity of 54.0% (53.4% to 54.5%) when combined with non-referable retinopathy. CONCLUSION: The algorithm demonstrated safe levels of sensitivity for high-risk retinopathy in a real-world screening service, with specificity that could halve the workload for human graders. AI machine learning and deep learning algorithms such as this can provide clinically equivalent, rapid detection of retinopathy, particularly in settings where a trained workforce is unavailable or where large-scale and rapid results are needed.

Association of Retinopathy and Retinal Microvascular Abnormalities With Stroke and Cerebrovascular Disease.

BACKGROUND AND PURPOSE: Abnormalities of the retinal circulation may be associated with cerebrovascular disease. We investigated associations between retinal microvascular abnormalities and (1) strokes and subclinical cerebral infarcts and (2) cerebral white matter lesions in a UK-based triethnic population-based cohort. METHODS: A total of 1185 participants (age, 68.8±6.1 years; 77% men) underwent retinal imaging and cerebral magnetic resonance imaging. Cerebral infarcts and white matter hyperintensities were identified on magnetic resonance imaging, retinopathy was graded, and retinal vessels were measured. RESULTS: Higher retinopathy grade (odds ratio [OR], 1.40 [95% confidence interval (95% CI), 1.16-1.70]), narrower arteriolar diameter (OR, 0.98 [95% CI, 0.97-0.99]), fewer symmetrical arteriolar bifurcations (OR, 0.84 [95% CI, 0.75-0.95]), higher arteriolar optimality deviation (OR, 1.16 [95% CI, 1.00-1.34]), and more tortuous venules (OR, 1.20 [95% CI, 1.09-1.32]) were associated with strokes/infarcts and white matter hyperintensities. Associations with quantitative retinal microvascular measures were independent of retinopathy. CONCLUSIONS: Abnormalities of the retinal microvasculature are independently associated with stroke, cerebral infarcts, and white matter lesions.

TOSCA-Imaging--developing Internet based image processing software for screening and diagnosis of diabetic retinopathy.

The primary aim of TOSCA-Imaging, which was a part of the TOSCA Project financed by EU's Fifth Framework IST Programme, was to develop Internet based software and image data bases for screening and diagnosis of diabetic retinopathy and implementing it into a real life situation. The work consisted of: 1) Construction of an Internet based communication platform for transmitting and analyzing retinal images. 2) Implementation of routines for detecting the first microaneurysm (transition from normal to pathologic), detecting patients needing referral for treatment (presence of venous beading or hard exudates near the fovea), and for serial analysis (image alignment). 3) Construction of a reference image data base. A preliminary validation showed that decisions that depended on a precise detection of individual lesions, e.g. the detection of normality, had a sensitivity and specificity of around 80%, whereas decisions that depended on the detection of lesion patterns, e.g. clinically significant macular oedema, had a sensitivity and specificity of more than 95%. Validation of the reference image data base by double grading by two expert graders suggested a sensitivity and a specificity of just below 90% for any lesion and of more than 95% for predicting the overall retinopathy grade. TOSCA-Imaging has succeeded in its primary aim of developing Internet based software and implementing it into a real life situation, integrating work within image processing done in four different European countries (England, Germany, Ireland, and Denmark) to be accessed from one Internet web site.