ABSTRACT
Objective:To compare the consistency and difference of non-mydriatic two-field 45° ultra-wide field Optos and Clarus500 fundus imaging in a large-scale diabetic retinopathy (DR) screening.Methods:A diagnostic methodology study. From November 2020 to August 2021, 526 eyes of 277 patients with type 2 diabetes who diagnosed in Department of Ophthalmology, Henan Provincial People's Hospital were included in the study. Among them, there were 175 males with 328 eyes and 102 females with 198 eyes; the age was 53±10 years old. The same experienced technician performed the non-mydriatic dual-field 45° fundus imaging and the non-mydriatic ultra-wide-angle imaging system Optos, Clarus500 single-field fundus imaging examination on the patient on the same day, and obtained the dual-field 45° fundus image and Optos, Clarus500 single-field fundus image. The Optos and Clarus500 single-field fundus images in the same area as the dual-field 45° fundus image were captured by Photoshop software, and the Optos and Clarus500 dual-field fundus images were obtained. Subsequently, two experienced ophthalmologists performed interpretation and DR grading of the 5 groups of images, respectively. Images with inconsistent grading results were interpreted by a third ophthalmologist and used as the final grading result. In order to avoid the mydriatic dual-field 45° imaging interpretation results as the standard, the consistency and detection rate difference of the two ultra-wide-angle imaging systems in the rapid DR screening results were evaluated. The weighted Kappa ( κ) test was used to analyze the consistency of DR diagnosis between dual-field 45° fundus imaging and Optos and Clarus500 fundus imaging; χ2 test was used to compare the detection rates of DR between different imaging systems. Results:Compared with the dual-field 45° fundus image, the Clarus500 single-field had a higher DR detection rate ( χ2=24.965, P<0.001), and the Optos dual-field fundus image had a lower DR detection rate ( χ2=49.559, P<0.001). Compared with the DR detection rate of dual-field 45° fundus image, Optos single-field fundus image, Clarus500 double-field fundus image had no significant difference ( χ2=2.572, 0.649; P=0.109, 0.421). Compared with Optos, Clarus500 single-field and dual-field fundus images DR detection rate, the difference was statistically significant ( χ2=43.214, 61.216; P<0.001). Consistency assessment of DR grading results: dual-field 45° fundus images and Clarus500 dual-field fundus images ( κ value=0.932, 95% confidence interval ( CI) 0.907-0.956) were highly consistent; dual-field 45° fundus images and Optos single-field fundus images [ κ value=0.474, 95% CI 0.417-0.532], Optos dual-field fundus image ( κ value=0.495, 95% CI 0.438-0.551), Optos dual-field fundus image ( κ value=0.495, 95% CI 0.438-0.551) and Clarus500 dual-field fundus image ( κ value=0.452, 95% CI 0.395-0.506) were moderately consistent; dual-field 45°fundus images and Clarus500 single-field fundus images ( κ value=0.354, 95% CI 0.303-0.403) and Optos single-field fundus images and Clarus500 single-field fundus images ( κ value=0.347, 95% CI 0.287-0.393) showed general agreement. Conclusions:Compared with Optos dual-field fundus image, dual-field 45°fundus image and Clarus500 dual-field fundus image have high consistency in the grading results of DR rapid screening. Compared with Optos single-field fundus image, the detection rate of the DR of Clarus500 single-field fundus image is higher.