Purchase this article with an account.
M. K. Smolek, S. Vujosevic, S. Piermarocchi, E. Midena, T. Peto, M. De Luca, E. Grisan, A. Ruggeri, S. D. Klyce; An Expert System for Diabetic Retinopathy Screening With a Non-Mydriatic, Operator-Free Fundus Camera. Invest. Ophthalmol. Vis. Sci. 2008;49(13):2725.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Worldwide, 240 million people have diabetes with 50% unaware of their condition. An estimated 2.5 million have diabetic retinopathy (DR), making it the leading cause of adult blindness. Fundus photography reading centers are over-burdened with the growing number of DR cases to review. This study describes an artificial intelligence (AI) / camera system that rapidly screens for DR.
Subjects were recruited from the patient population at the University of Padova Diabetes Clinic. A total of 130 non-DR and 11 DR eyes were selected for analysis. Non-DR included normal subjects and diabetic subjects with normal retinas or non-visually threatening (VT) disease. DR included only cases of VT disease. Cases in which retinal imaging was impossible (e.g., severe cataracts) were excluded from the study. A Nidek Orion non-mydriatic automated fundus camera (Nidek Technologies, srl, Padova, Italy) recorded 5 overlapping fundus fields that were combined into a single 45° mosaic image. Canon fundus photographs were also acquired at the same clinic visit using a 3-field modified ETDRS pattern under non-dilated and dilated conditions. The Canon images were later screened at the Moorfields Eye Hospital Reading Center to serve as a comparative gold standard. The Orion mosaic images were processed for contrast enhancement and passed through an analysis routine that extracted the location, size, and shape of bright and dark blobs. This information was passed to an expert system that determined if DR was present and if it was VT or non-VT based on location and size. Only VT results were flagged for DR output. Results of the expert system were then compared to the gold standard result using the same criteria.
The expert system had 2 false positive errors and 1 false negative error for 98.2% overall accuracy, 90.9% sensitivity, and 98.7% specificity when screening for VT DR on non-dilated eyes.
This expert system is an effective step toward mass screening of subjects with a risk for VT DR. The entire AI process can be contained within the Orion instrument, where it will provide rapid analysis and avoid concerns about data integrity and patient privacy that are associated with telemedicine methods. The expert system distinguishes referrable VT DR cases, thus saving time and improving the management of this blinding eye disease.
This PDF is available to Subscribers Only