EyeArt v2.2.0

EyeArt v2.2.0 is an artificial intelligence-powered software system designed to autonomously analyze retinal fundus images for the early detection of more than mild and vision-threatening diabetic retinopathy in adult diabetic patients. Its methodology relies on machine learning algorithms processing images from compatible Canon and Topcon fundus cameras, with clinical studies validating its high sensitivity and specificity for screening these conditions.

EyeArt v2.2.0 is a software as a medical device (SaMD) designed for healthcare providers to automatically detect more than mild diabetic retinopathy (mtmDR) and vision-threatening diabetic retinopathy (vtDR) in adults diagnosed with diabetes who have not been previously diagnosed with diabetic retinopathy. The system operates using artificial intelligence, specifically an ensemble of machine learning and deep learning algorithms within its Analysis Computation Engine, to analyze retinal fundus images; it comprises a Client software for image transfer and results display (including live image quality feedback), a secure Server for data handling, and the remote Analysis Computation Engine, and is indicated for use with Canon CR-2 AF, Canon CR-2 Plus AF, and Topcon NW400 cameras, requiring two specific 45-degree field-of-view images per eye. Clinical performance for EyeArt v2.2.0, particularly with the newly supported Topcon NW400 camera, was validated through a prospective multi-center study demonstrating high diagnostic accuracy (e.g., for Topcon NW400, mtmDR sensitivity 94.4%/specificity 91.1%; vtDR sensitivity 96.8%/specificity 91.6%), high agreement between camera models, and excellent intra-operator repeatability and inter-operator reproducibility. A retrospective study comparing v2.2.0 to its predicate (v2.1.0) confirmed these findings, showing improved imageability (99.1% final submission with v2.2.0 versus 97.3% with v2.1.0) and maintained high sensitivity and specificity, reflecting technological enhancements such as an updated image quality assessment module, support for the additional camera, live image quality feedback, and an option to retry imaging after pupil dilation.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the 'updated image quality assessment module' in EyeArt v2.2.0 differ algorithmically from v2.1.0, and what specific image artifacts does it better handle to achieve higher imageability?; What is the average image analysis processing time for EyeArt v2.2.0 from successful image acquisition to the display of results, and does this time vary significantly between the supported camera models (Canon CR-2 AF/Plus AF, Topcon NW400)?; The 'live image quality feedback' is a key new feature; what specific image parameters are assessed in real-time, and what is the typical operator learning curve or observed improvement in image capture success rates attributable to this feature?; For the small percentage of participant eyes (approx. 0.9-1.1%) whose images remained unanalyzable by EyeArt v2.2.0 even after pupil dilation, what were the predominant reasons, and are there ongoing developments to further reduce this ungradable rate?; How does EyeArt v2.2.0's performance (sensitivity, specificity, imageability) vary across different demographic subgroups, such as varying durations of diabetes, HbA1c levels, or ethnic backgrounds, beyond the overall cohort statistics presented?