Diagnostic Accuracy of Community-Based Diabetic Retinopathy Screening With an Offline Artificial Intelligence System on a Smartphone

There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

Abstract

<div class="section"> <a class="named-anchor" id="ab-eoi190055-1">  </a> <h5 class="section-title" id="d8278802e265">Question</h5> <p id="d8278802e267">What is the performance of an offline, automated artificial intelligence system of analysis to detect referable diabetic retinopathy on images taken by a health worker on a smartphone-based, nonmydriatic retinal camera? </p> </div><div class="section"> <a class="named-anchor" id="ab-eoi190055-2">  </a> <h5 class="section-title" id="d8278802e270">Finding</h5> <p id="d8278802e272">In this cross-sectional study, fundus images from 213 study participants were subjected to offline, automated analysis. The sensitivity and specificity of the analysis to diagnose referable diabetic retinopathy were 100.0% and 88.4%, respectively, and the sensitivity and specificity for any diabetic retinopathy were 85.2% and 92.0%, respectively. </p> </div><div class="section"> <a class="named-anchor" id="ab-eoi190055-3">  </a> <h5 class="section-title" id="d8278802e275">Meaning</h5> <p id="d8278802e277">This study suggests these methods might be used to screen for referable diabetic retinopathy using offline artificial intelligence and a smartphone-based, nonmydriatic retinal imaging system. </p> </div><p class="first" id="d8278802e280">This cross-sectional study compares the diagnostic accuracy of a smartphone-based artificial intelligence system vs ophthalmologist judgement in patients with referable diabetic retinopathy or any diabetic retinopathy in Mumbai, India. </p><div class="section"> <a class="named-anchor" id="ab-eoi190055-4">  </a> <h5 class="section-title" id="d8278802e284">Importance</h5> <p id="d8278802e286">Offline automated analysis of retinal images on a smartphone may be a cost-effective and scalable method of screening for diabetic retinopathy; however, to our knowledge, assessment of such an artificial intelligence (AI) system is lacking. </p> </div><div class="section"> <a class="named-anchor" id="ab-eoi190055-5">  </a> <h5 class="section-title" id="d8278802e289">Objective</h5> <p id="d8278802e291">To evaluate the performance of Medios AI (Remidio), a proprietary, offline, smartphone-based, automated system of analysis of retinal images, to detect referable diabetic retinopathy (RDR) in images taken by a minimally trained health care worker with Remidio Non-Mydriatic Fundus on Phone, a smartphone-based, nonmydriatic retinal camera. Referable diabetic retinopathy is defined as any retinopathy more severe than mild diabetic retinopathy, with or without diabetic macular edema. </p> </div><div class="section"> <a class="named-anchor" id="ab-eoi190055-6">  </a> <h5 class="section-title" id="d8278802e294">Design, Setting, and Participants</h5> <p id="d8278802e296">This prospective, cross-sectional, population-based study took place from August 2018 to September 2018. Patients with diabetes mellitus who visited various dispensaries administered by the Municipal Corporation of Greater Mumbai in Mumbai, India, on a particular day were included. </p> </div><div class="section"> <a class="named-anchor" id="ab-eoi190055-7">  </a> <h5 class="section-title" id="d8278802e299">Interventions</h5> <p id="d8278802e301">Three fields of the fundus (the posterior pole, nasal, and temporal fields) were photographed. The images were analyzed by an ophthalmologist and the AI system. </p> </div><div class="section"> <a class="named-anchor" id="ab-eoi190055-8">  </a> <h5 class="section-title" id="d8278802e304">Main Outcomes and Measures</h5> <p id="d8278802e306">To evaluate the sensitivity and specificity of the offline automated analysis system in detecting referable diabetic retinopathy on images taken on the smartphone-based, nonmydriatic retinal imaging system by a health worker. </p> </div><div class="section"> <a class="named-anchor" id="ab-eoi190055-9">  </a> <h5 class="section-title" id="d8278802e309">Results</h5> <p id="d8278802e311">Of 255 patients seen in the dispensaries, 231 patients (90.6%) consented to diabetic retinopathy screening. The major reasons for not participating were unwillingness to wait for screening and the blurring of vision that would occur after dilation. Images from 18 patients were deemed ungradable by the ophthalmologist and hence were excluded. In the remaining participants (110 female patients [51.6%] and 103 male patients [48.4%]; mean [SD] age, 53.1 [10.3] years), the sensitivity and specificity of the offline AI system in diagnosing referable diabetic retinopathy were 100.0% (95% CI, 78.2%-100.0%) and 88.4% (95% CI, 83.2%-92.5%), respectively, and in diagnosing any diabetic retinopathy were 85.2% (95% CI, 66.3%-95.8%) and 92.0% (95% CI, 97.1%-95.4%), respectively, compared with ophthalmologist grading using the same images. </p> </div><div class="section"> <a class="named-anchor" id="ab-eoi190055-10">  </a> <h5 class="section-title" id="d8278802e314">Conclusions and Relevance</h5> <p id="d8278802e316">These pilot study results show promise in the use of an offline AI system in community screening for referable diabetic retinopathy with a smartphone-based fundus camera. The use of AI would enable screening for referable diabetic retinopathy in remote areas where services of an ophthalmologist are unavailable. This study was done on patients with diabetes who were visiting a dispensary that provides curative services to the population at the primary level. A study with a larger sample size may be needed to extend the results to general population screening, however. </p> </div>

Related collections

Most cited references 11

Record: found
Abstract: not found
Conference Proceedings: not found

Learning Deep Features for Discriminative Localization

Aditya Khosla, Antonio S. Torralba, Aude Oliva … (2016)

0 comments Cited 737 times – based on 0 reviews

Bookmark

Record: found
Abstract: found
Article: found

Is Open Access

The English National Screening Programme for diabetic retinopathy 2003–2016

Peter H Scanlon (2017)

The aim of the English NHS Diabetic Eye Screening Programme is to reduce the risk of sight loss amongst people with diabetes by the prompt identification and effective treatment if necessary of sight-threatening diabetic retinopathy, at the appropriate stage during the disease process. In order to achieve the delivery of evidence-based, population-based screening programmes, it was recognised that certain key components were required. It is necessary to identify the eligible population in order to deliver the programme to the maximum number of people with diabetes. The programme is delivered and supported by suitably trained, competent, and qualified, clinical and non-clinical staff who participate in recognised ongoing Continuous Professional Development and Quality Assurance schemes. There is an appropriate referral route for those with screen-positive disease for ophthalmology treatment and for assessment of the retinal status in those with poor-quality images. Appropriate assessment of control of their diabetes is also important in those who are screen positive. Audit and internal and external quality assurance schemes are embedded in the service. In England, two-field mydriatic digital photographic screening is offered annually to all people with diabetes aged 12 years and over. The programme commenced in 2003 and reached population coverage across the whole of England by 2008. Increasing uptake has been achieved and the current annual uptake of the programme in 2015–16 is 82.8% when 2.59 million people with diabetes were offered screening and 2.14 million were screened. The benefit of the programme is that, in England, diabetic retinopathy/maculopathy is no longer the leading cause of certifiable blindness in the working age group.

0 comments Cited 182 times – based on 0 reviews      Review now

Bookmark

Record: found
Abstract: found
Article: not found

Screening for diabetic retinopathy: 1 and 3 nonmydriatic 45-degree digital fundus photographs vs 7 standard early treatment diabetic retinopathy study fields.

Stela Vujosevic, Elisa Benetti, Francesca Massignan … (2009)

To evaluate if simple- or multiple-field digital color nonmydriatic (NM) retinal images can replace 7 standard stereoscopic fundus photographs in the screening of diabetic retinopathy (DR). Prospective, masked, comparative case series. One hundred and eight eyes of 55 diabetics were studied to determine single lesions and to grade clinical levels of DR and diabetic macular edema (DME) using both 1 and 3 NM digital color retinal images compared with the Early Treatment Diabetic Retinopathy Study (ETDRS) 7 standard 35-mm stereoscopic color fundus photographs (7F-ETDRS). All eyes underwent NM 45-degree field images of 1 central field (1F-NM), NM 45-degree field images of 3 fields (3F-NM), and, after pupil dilatation, 30-degree 7F-ETDRS photography. Images were analyzed by 2 independent, masked retinal specialists (S.V. and E.B.), lesion-by-lesion according to the ETDRS protocol and for clinical severity level of DR and DME according to the international classification of DR. Using 7F-ETDRS as the gold standard, agreement was substantial for grading clinical levels of DR and DME (kappa = 0.69 and kappa = 0.75) vs 3F-NM; moderate for DR level (kappa = 0.56) and substantial for DME (kappa = 0.66) vs 1F-NM; almost perfect for detecting presence or absence of DR (kappa = 0.88) vs both 1F-NM and 3F-NM; and almost perfect for presence or absence of DME (kappa = 0.97) vs 3F-NM and substantial (kappa = 0.75) vs 1F-NM. Sensitivity and specificity for detecting referable levels of DR were 82% and 92%, respectively, for 3F-NM and 71% and 96%, respectively, for 1F-NM. Three color 45-degree NM fundus fields may be an effective tool in a screening setting to determine critical levels of DR and DME for prompt specialist referral. One central 45-degree image is sufficient to determine absence or presence of DR and DME, but not for grading it.