Validation Study of RETINA-AI Galaxy™ v2.0, an Automated Diabetic Retinopathy Screening Device

Diabetes affects over 37 million Americans and over 530 million people globally. Each diabetic patient needs at least one retinal exam per year starting immediately at the time of diagnosis if they have Type II diabetes (and starting at 5th year after disease onset if they have Type I diabetes). However, majority of diabetic patients do not get their eye exam due to multiple prohibitive factors such as cost, transportation, difficulty of taking time off from work, and inconvenience, amongst others. As a result, diabetes is the most common cause of visual impairment and blindness in working age adults in the United States and globally. Early detection via effective screening can prevent diabetes-related blindness. However, there are multiple barriers to screening. This prompted the development of RETINA-AI Galaxy™ v2.0, an automated Software as a Medical Device that screens for diabetic retinopathy in the primary care setting. This observational study was designed to validate the safety and efficacy of the RETINA-AI Galaxy™ Software-as-a-Medical-Device.

This study was a prospective, multicenter, observational study to assess the safety and efficacy of the RETINA-AI Galaxy™ v2.0 ("Galaxy™") in screening for diabetic retinopathy in the primary care setting. The study design conformed to an Intent to Screen (ITS) paradigm. The Galaxy™ is a Software-as-a-Medical-Device designed to analyze digital color fundus photographs and rapidly screen for diabetic retinopathy in the primary care setting. Subjects who met eligibility criteria were recruited from Sites in the United States staffed by primary care providers. Eligibility was assessed and informed consent was obtained, after which digital color fundus photographs were taken using U.S. Food and Drug Administration (FDA) cleared non-mydriatic fundus cameras, by an operator using the Galaxy photography manual. There were a total of 3 non-mydriatic robotic screening cameras used in the AI system protocol part of the study. There was a dedicated validation camera used in the Validation Reading Center Protocol part of the study at each Site. Primary care clinical staff (e.g. medical assistant) with no prior professional ophthalmic photography experience and only a 4 hour training operated the RETINA-AI Galaxy v2.0 device and the screening cameras. The Retina Reading Center (RRC)- certified professional ophthalmic photographers operated the validation fundus cameras according the the 4W-D stereo protocol, and obtained OCT images of the macula.