What are the FDA (Food and Drug Administration) approved artificial intelligence (AI) tools for ophthalmology?

From the Guidelines

The FDA has approved several AI tools for ophthalmology, including IDx-DR, EyeArt, and Pegasus, which assist in diagnosing and managing eye conditions such as diabetic retinopathy and retinal diseases. These AI tools work by analyzing medical images using deep learning algorithms trained on thousands of examples to identify patterns associated with specific eye conditions 1. They help address the shortage of eye care specialists, especially in underserved areas, by providing quick and accurate screening results. According to a study published in 2022, AI systems that detect more than mild diabetic retinopathy and diabetic macular edema, authorized for use by the U.S. Food and Drug Administration (FDA), represent an alternative to traditional screening approaches 1.

Key Features of FDA-Approved AI Tools

• IDx-DR: analyzes retinal photos and provides a screening decision for patients with diabetes
• EyeArt: detects diabetic retinopathy
• Pegasus: detects retinal diseases from OCT scans
• These systems typically require standard ophthalmic imaging equipment and can be integrated into existing clinical workflows

Benefits of AI Tools in Ophthalmology

• Quick and accurate screening results
• Help address the shortage of eye care specialists, especially in underserved areas
• Complement comprehensive eye examinations by healthcare professionals
• Can be used to detect various ocular conditions, including diabetic retinopathy, glaucoma, and age-related macular degeneration 1

Operational Workflow of AI Systems

• Registration: ability of an AI system to detect types of ocular images, field of view, and gradeability
• Diagnosis and consultation: preprocessing, analysis, and generation of diagnosis
• Output: classifications, sometimes binary but sometimes multi-class outputs 1

From the Research

FDA Approved AI Tools for Ophthalmology

• The FDA has approved AI tools such as IDx-DR (IDx Technologies Inc, USA) for autonomous detection of referable diabetic retinopathy 2.
• Other AI systems like RetCAD (Thirona, The Netherlands) have shown high sensitivity and specificity in detecting age-related macular degeneration (AMD) and diabetic retinopathy (DR) simultaneously based on fundus photos 3.
• Deep learning frameworks have also been developed for the diagnosis of eye diseases such as DR, AMD, and glaucoma based on structural and angiographic OCT scans, achieving high area under the curve (AUC) values 4.

Applications of AI in Ophthalmology

• AI can aid in the detection of diabetic retinopathy, age-related macular degeneration, and glaucoma through automated screening and diagnosis 2, 3, 4.
• AI algorithms can analyze retinal images to accurately identify lesions and detect damage to the optic nerve head 2.
• AI can also quantify drusen and diagnose disease severity from both color fundus and optical coherence tomography images 2.

Performance of AI Tools

• The RetCAD system achieved an area under the ROC curve (Az) of 0.926 with a sensitivity of 84.6% at a specificity of 84.0% for image quality 3.
• The deep learning framework achieved AUC values of 0.95,0.98, and 0.91 for the diagnosis of DR, AMD, and glaucoma, respectively 4.
• A systematic review and meta-analysis found that AI had a pooled AUROC of 0.983, sensitivity of 0.88, specificity of 0.90, and diagnostic odds ratio of 275.27 for detecting AMD in color fundus photographs 5.