Comparison of quantitative assessment and efficiency of diabetic retinopathy diagnosis using ETDRS seven-field imaging and two ultra-widefield imaging.

Abstract

This study compared the efficiency of diabetic retinopathy (DR) diagnosis and differences in the relative visible retinal area among the Early Treatment Diabetic Retinopathy Study (ETDRS) seven-field, ultra-widefield (UWF)-Optos, and UWF-Clarus fundus imaging methods. This was a prospective and clinic-based comparative study. All patients underwent three fundus examinations, and all images were graded using the ETDRS severity scale. We compared and analysed the agreement of DR severity and the relative visible retinal area among the three fundus examination methods, and the number and type of lesions outside the ETDRS seven-field (peripheral lesions) between the two UWF imaging methods. A total of 202 patients (386 eyes) were included. Weighted kappa for the agreement between ETDRS seven-field and blinded Optos images was 0.485; between ETDRS seven-field and blinded Clarus images, 0.924; and between blinded Optos and Clarus images, 0.461. Blinded Clarus showed excellent performance when a ETDRS scale was used for grading the images. The relative visible retinal area for ETDRS seven-field images was 195 ± 28 disc area (DA); single Optos images, 371 ± 69 DA; single Clarus images, 261 ± 65 DA; two-montage Clarus images, 462 ± 112 DA; and four-montage Clarus images, 598 ± 139 DA. The relative visible retinal area was statistically significant between any two of the imaging systems used. In total, 2015 and 4200 peripheral lesions were detected in single Optos and Clarus images, respectively (P < 0.001). These peripheral lesions on two UWF images suggested a more severe DR level in approximately 10% and 12% of eyes, respectively. UWF-Clarus fundus imaging offers a suitable assessment approach for DR severity; it could improve DR diagnosis and has the potential to replace ETDRS seven-field imaging after additional clinical trials.