Evaluation of The Performance of Retinow AI Software

The main purpose of this clinical study is to evaluate the effectiveness and reliability of Retinow AI software in the diagnosis of common eye diseases such as diabetic retinopathy, glaucoma and macular degeneration. Retinow AI is a cloud-based artificial intelligence software that aims to detect diabetic retinopathy, glaucoma and macular degeneration diseases through fundus photographs. The software stands out with its ability to detect disease symptoms at an early stage and accelerate the diagnosis process. In addition, it is claimed that this software, which has reached a 90% accuracy rate during pre-clinical validation studies, can achieve similar results to the diagnostic accuracy of specialist physicians. This study examines the usability of Retinow AI software by both specialist and non-specialist physicians and its potential to save time in diagnostic processes. It is anticipated that the software can improve patient management, reduce costs and increase the efficiency of general healthcare services by accelerating the diagnosis of eye diseases. Certain eligibility criteria have been defined for the subjects and users to be examined within the scope of the study. These criteria are designed to reliably evaluate the performance of Retinow AI software. Retinow AI is designed for use by healthcare providers. The user must have sufficient understanding of the language in which the user manual was prepared.

Primary Objective:

The primary objective is to evaluate the Retinow AI software's ability to diagnose diabetic retinopathy, glaucoma, and macular degeneration diseases with high accuracy through fundus photographs. The software's performance was compared with diagnoses made by specialist physicians, and accuracy, sensitivity, and specificity metrics were measured. The hypothesis that Retinow AI can achieve 90% accuracy was tested. In this context, the Retinow AI's ability to consistently identify the same disease symptoms in different fundus images was evaluated by analyzing false positive and false negative results.

Primary Hypothesis:

It is hypothesized that Retinow AI software can diagnose eye diseases such as diabetic retinopathy, glaucoma, and macular degeneration at an early stage through fundus photographs with 90% accuracy. It is anticipated that the software can achieve similar sensitivity and specificity rates to the evaluations of specialist physicians in the diagnosis of these diseases.

This clinical study is a single-center, observational, prospective, and cross-sectional study conducted at Ankara Bilkent City Hospital to determine the primary endpoints of sensitivity, specificity, and accuracy of the Retinow AI device for diabetic retinopathy, glaucoma, and macular degeneration.

The clinical trial, clinical trial plan, clinical trial results report were conducted in accordance with the ethical principles of the Declaration of Helsinki, the principles of SS-EN ISO 14155:2020, and the current national and international regulations governing this clinical trial. A signed Declaration of Helsinki was obtained from all participants in the clinical trial.

The study included healthy participants who did not have diabetic retinopathy, glaucoma, macular degeneration, or any eye disease (no pathological findings in the retina). The study population consisted of 940 participants aged 18 years and older who applied to an ophthalmologist for an eye examination to detect diabetic retinopathy, macular degeneration, and glaucoma. 25 of the participants were excluded from the study because they did not meet the inclusion criteria. The study included 153 participants for Diabetic Retinopathy, 153 for Glaucoma, and 153 for Macular Degeneration. A total of 456 participants, 152 from each disease group, were healthy participants with no pathological findings in their retinas. A total of 915 participants who completed all procedures were included in this study. Pregnant women were not included in the study. Pregnancy does not have a long-term effect on diabetic retinopathy. Even if retinopathy progresses during pregnancy, it regresses after delivery. Retinow AI is not intended for use in patients with gestational diabetes because diabetic retinopathy can progress very rapidly during pregnancy. Retinow AI is not intended to evaluate rapidly progressing diabetic retinopathy. The software is designed to detect diabetic retinopathy, glaucoma, and macular degeneration only and should not be used to detect other diseases or conditions.Methods and Tools Used in the Study:

• In order to diagnose diabetic retinopathy, glaucoma and macular degeneration, analyze the retina of healthy individuals and enable specialist physicians to diagnose diseases from fundus images, a Canon Europe BV brand color video 3CCD camera mounted on a Topcon TRC-NW6 nonmydriatic fundus camera (Topcon USA, Inc.) with a 45° field of view and centered on the fovea was used during routine eye examinations.
• Retinow AI software: Retinow software was used by retina specialists to evaluate fundus images and compare results for clinical validation of the Retinow AI device. Retinow AI is an artificial intelligence-based software that diagnoses eye diseases by analyzing fundus images. The analysis results of the software were periodically compared with the results of specialist ophthalmologists. Since the latest version of the Retinow AI software was used, no changes were made to the software throughout the clinical study.