--- title: "FDA-Cleared AI for Diabetic Retinopathy Screening: How IDx-DR (Now LumineticsCore) Works" slug: fda-ai-diabetic-retinopathy-screening-idx-dr category: health category_label: "Health" author: "BrainWavePost Staff" date: 2026-06-08 tags: ["FDA", "diabetic retinopathy", "AI", "healthcare", "screening", "diabetes", "LumineticsCore"] read_time_minutes: 7 canonical_url: https://brainwavepost.com/article/fda-ai-diabetic-retinopathy-screening-idx-dr source: BrainWavePost --- # FDA-Cleared AI for Diabetic Retinopathy Screening: How IDx-DR (Now LumineticsCore) Works *Health · 2026-06-08 · BrainWavePost Staff · 7 min read* > In 2018 the FDA authorized the first fully autonomous AI diagnostic system in medicine — IDx-DR, now called LumineticsCore — to detect diabetic retinopathy in primary care. Here is a sourced, neutral guide to how it works and what the pivotal trial showed. > **How this article is sourced** _(info)_ > > Every factual claim below is taken directly from the U.S. Food and Drug Administration (FDA), the peer-reviewed pivotal trial published in npj Digital Medicine (Nature Portfolio), the earlier algorithm study published in JAMA Ophthalmology, and the manufacturer Digital Diagnostics (formerly IDx). This is general health information, not medical advice. [1][2][3][4][5] On 11 April 2018, the U.S. Food and Drug Administration announced it had permitted marketing of IDx-DR — an artificial-intelligence-based diagnostic system designed to detect diabetic retinopathy in adults with diabetes. [1] The FDA described it as the first medical device to use AI to detect greater than a mild level of diabetic retinopathy without requiring a clinician to interpret the images. [1] ## What is diabetic retinopathy? Diabetic retinopathy is a complication of diabetes that affects the blood vessels of the retina. According to the CDC, it is the leading cause of blindness among working-age adults in the United States. [6] The condition often has no early symptoms, which is why regular screening is recommended for people with diabetes. [6] ## What the FDA authorized The FDA granted IDx-DR a De Novo classification request (DEN180001) on 12 January 2018, establishing a new regulatory classification under 21 CFR 886.1100 for retinal diagnostic software devices. [2][3] The device is a Class II prescription software system with product code PIB. [2] According to the FDA's indications for use, LumineticsCore (formerly IDx-DR) is intended for use by healthcare providers to automatically detect more than mild diabetic retinopathy and diabetic macular edema in adults aged 22 years or older who have been diagnosed with diabetes and have not been previously diagnosed with diabetic retinopathy. [4] The system is designed for use with the Topcon NW400 retinal camera. [4] > **What 'autonomous' means here** _(note)_ > > The term 'autonomous' means the AI makes a diagnostic determination without a human clinician reviewing the retinal images. The primary care operator captures the images; the software analyzes them and produces a result at the point of care. [2][5] ## The pivotal clinical trial The FDA authorization was based on a pivotal trial of 900 participants enrolled at 10 primary care clinic sites, with results published in npj Digital Medicine (Nature Portfolio). [2][5] The study compared the AI system's output against a gold-standard reference: grading by the Wisconsin Fundus Photograph Reading Center using widefield stereoscopic photography and macular Optical Coherence Tomography (OCT). [5] Among the 819 fully analyzable participants, 198 (23.8%) had more than mild diabetic retinopathy (mtmDR) according to the reference standard. [5] The trial was conducted under an intent-to-screen design, with existing primary care staff operating the AI system after standardized training. [5] - **87.2%** — Sensitivity for mtmDR (95% CI, 81.8–91.2%) [5] - **90.7%** — Specificity for mtmDR (95% CI, 88.3–92.7%) [5] - **96.1%** — Imageability rate (95% CI, 94.6–97.3%) [5] - **97.4%** — Sensitivity for vision-threatening DR [5] ## What the trial results mean The AI system exceeded all pre-specified superiority endpoints. [5] Sensitivity of 87.2% means the system correctly identified approximately 87 out of every 100 people who truly had more than mild diabetic retinopathy. Specificity of 90.7% means it correctly ruled out the condition in approximately 91 out of every 100 people who did not have it. [5] The npj Digital Medicine study also noted that the system achieved 97.4% sensitivity for vision-threatening diabetic retinopathy — the most severe cases requiring urgent specialist referral. [5] Logistic regression did not identify any significant effects of age, sex, race, ethnicity, or study site on the system's sensitivity or specificity, suggesting consistent performance across demographic groups. [5] ## How the system is used today The device, now marketed as LumineticsCore by Digital Diagnostics, is designed to be operated by trained non-specialist staff in primary care settings. [4] After the retinal images are captured, the algorithm produces a diagnostic result within approximately 30 seconds. [4] If the system detects more than mild diabetic retinopathy, the patient is referred to an eye care specialist for further evaluation and possible treatment. If the system does not detect the condition, the patient can be rescreened at the interval recommended by standard diabetes care guidelines. [4] The system is reimbursable under CPT code 92229. [4] ## The research behind the algorithm The underlying image-analysis algorithm has its roots in earlier research published in JAMA Ophthalmology in 2013 by Abramoff et al., which evaluated an automated analysis system (then called the Iowa Detection Program) for detecting referable diabetic retinopathy. [3] That study demonstrated high diagnostic accuracy on a separate dataset, providing foundational evidence for the later FDA-authorized system. [3] ## Why this matters for diabetes care According to the npj Digital Medicine study, less than 50% of patients with diabetes adhere to recommended eye exam schedules, and adherence has not increased over the preceding 15 years despite large-scale public health efforts. [5] By bringing autonomous screening into primary care offices — where patients already go for diabetes management — the system aims to close this care gap and catch retinopathy earlier, when treatment is most effective. [5] > **Not a substitute for comprehensive eye care** _(note)_ > > LumineticsCore screens for diabetic retinopathy and diabetic macular edema only. It does not detect all eye diseases, and it is not intended for patients who already have a diagnosis of diabetic retinopathy, who are pregnant, who have visual symptoms, or who have previously received retinal laser treatment or eye injections. [4] Patients should continue to follow their healthcare provider's recommendations for overall eye care. ## References (clickable) - [1] U.S. Food and Drug Administration — 'FDA permits marketing of artificial intelligence-based device to detect certain diabetes-related eye problems' (11 April 2018): https://www.fda.gov/news-events/press-announcements/fda-permits-marketing-artificial-intelligence-based-device-detect-certain-diabetes-related-eye - [2] U.S. Food and Drug Administration — De Novo Classification Request for IDx-DR (DEN180001) (12 January 2018): https://www.accessdata.fda.gov/cdrh_docs/reviews/DEN180001.pdf - [3] Abràmoff MD et al. — 'Automated Analysis of Retinal Images for Detection of Referable Diabetic Retinopathy' (JAMA Ophthalmology, 2013): https://jamanetwork.com/journals/jamaophthalmology/fullarticle/1668203 - [4] Digital Diagnostics — LumineticsCore Indications for Use – US: https://www.digitaldiagnostics.com/products/eye-disease/indications-for-use-us/ - [5] Abràmoff MD et al. — 'Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices' (npj Digital Medicine, 2018): https://www.nature.com/articles/s41746-018-0040-6 - [6] U.S. Centers for Disease Control and Prevention — Diabetic Retinopathy: https://www.cdc.gov/diabetes/diabetes-complications/diabetic-retinopathy.html --- _Canonical article: [https://brainwavepost.com/article/fda-ai-diabetic-retinopathy-screening-idx-dr](https://brainwavepost.com/article/fda-ai-diabetic-retinopathy-screening-idx-dr) — © BrainWavePost. Educational content; see the article page for full disclaimers._