Abstract
Purpose: :
To compare diabetic macular edema (DME) graded from monoscopic digital color retinal images to stereoscopic film color photosand stereoscopic color digital images.
Methods: :
Film (F) and digital (D) color images of 154 eyes of persons with diabetes stratified by severity level were taken with a Topcon 35o camera as 7 stereo fields (7SF) per DRS/ETDRS protocol. Monoscopic D images were created from stereo digital images by selecting the best of each stereo pair. Three certified graders independently evaluated macular edema status (according to ETDRS definitions) from F, and from mono and stereo D (2392 x 2048 pixel, uncompressed images viewed on calibrated 20" LCD monitors). Digital images were standardized for brightness, contrast and color-balance according to the AREDS2 image model, yielding film-like consistency.
Results: :
Graders classified DME status (present/absent) from F images as: any retinal thickening (RT) within ETDRS macular grid, 62 eyes; "Clinically Significant Macular Edema (CSME)," 48 eyes; and RT involving macular center, 40 eyes (15 probable, 25 definite).Comparison of mono D vs. stereo F gradings yielded the following:- For RT within grid, 93.5% agreement (unweighted Κ = 0.87, SEΚ = 0.04)- For CSME, 92.2% agreement (Κ = 0.81, SEΚ = 0.05), and- For RT at center, 86.4% agreement (Κ = 0.63, SEΚ = 0.07)Comparing stereo D vs. stereo F DME gradings revealed better agreement for RT at center (92.8%), but only marginally higher agreement for RT within grid and CSME.Comparison of mono Dwith stereo D showed similar agreement to that observed with mono D vs. stereo F.Differences between F and both mono and stereo D did not appear to be systematic, except that F detected more CSME than did either D variant - about 20% more cases.
Conclusions: :
Our DME evaluations from digital mono images were reasonably similar to those from film stereo photos and also from D stereo images. These results suggest that stereo effect, although very helpful, may not always be critical to DME evaluation.
Keywords: diabetic retinopathy • clinical (human) or epidemiologic studies: systems/equipment/techniques • imaging/image analysis: clinical