Abstract
Purpose: :
To compare gradings of diabetic retinopathy from film versus digital photographs obtained on a subset of DRCR.net participants.
Methods: :
The DRCR.net transition of film to digital study required simultaneous acquisition of 7 field standard color film fundus photos and either 7 field standard or 4 field wide angle digital images. Participants gave informed consent and the study had IRB approval. Participants were recruited from multiple clinical centers and across various DRCR.net protocols. Digital and film camera systems and photographers were certified to obtain images according to standard procedures. Images from the two systems were graded independently, for level of diabetic retinopathy by reading center graders in a masked fashion and classified according to the ETDRS retinopathy severity scale.
Results: :
153 eyes from 95 study participants had images submitted using both film and digital modes; 48 (29%) of the digital images used wide-angle settings. The retinopathy level on film and digital images, included 39 (23%) and 35 (21%) eyes no worse than moderate non-proliferative diabetic retinopathy (NPDR), 55 (33%) and 51 (30%) eyes with moderately severe or severe NPDR, 70 (42%) and 73 (43%) eyes with mild PDR or worse, and could not be graded in 4(2%) and 9 (5%) eyes. Retinopathy level agreed exactly in 74% and was within 1 step in 92% with a weighted kappa statistic of 0.80 (95% CI: 0.73-0.86). Sensitivity and specificity of digital versus film mode for detecting severe NPDR or PDR was 96% (95%CI: 84%-99%), and 92% (95%CI: 85%-96%), respectively.
Conclusions: :
Agreement was very good to excellent on level of diabetic retinopathy using stereoscopic photographs of digital images as compared with film images across a range of retinopathy levels among many different clinical sites participating in the DRCR Network. Advantages of digital images include the ability to review images immediately, perform retake images when quality is deficient, internet transmission of images and virtual storage of the source images.
Keywords: clinical (human) or epidemiologic studies: systems/equipment/techniques • diabetic retinopathy • imaging/image analysis: clinical