Purchase this article with an account.
H. K. Li, L. D. Hubbard, R. P. Danis, J. F. Florez-Arango, A. Esquivel, E. A. Krupinski; Evaluation of Diabetic Macular Edema From Multiple Digital Formats Compared to Film. Invest. Ophthalmol. Vis. Sci. 2010;51(13):4659.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Multiple digital imaging formats are available for telemedicine evaluation of sight-threatening diabetic retinopathy (DR). Their performance has not been systemically compared to the accepted standard Early Treatment Diabetic Retinopathy Study (ETDRS) color stereo film protocol. This study compares four digital variants (stereo uncompressed, stereo compressed, monoscopic and wide-angle mosaic) to ETDRS.
Film (F) and digital (D) color images of 152 eyes pre-selected for DR were taken with a Topcon 35 degree camera per ETDRS protocol. D images were also taken at 45 degree (Topcon TRC-NW6S) and auto-mosaicked with IMAGEnet 2000. Uncompressed 35 degree images were compressed at 37:1 using JPEG 2000. Monoscopic D images were created by selecting the better of each stereo D pair. Three certified graders independently determined diabetic macular edema (DME) according to ETDRS definitions. Custom software controlled grading order and timing to minimize bias/recall. Status of hard exudates (HE), retina thickening in the macula (RT), clinically significant macular edema (CSME) and retinal thickening at the macular center (RTMC) were defined by graders' median score.
McNemar test for overall bias of film results vs. all digital formats showed differences were not statistically significant for HE or RT. Differences were significant or near significant for CSME (p of Stereo D vs. F = 0.02, p of Compressed D vs. F = 0.02, Mono D vs. F = 0.05 and Wide-angle mosaic D vs. F = 0.02) and for RTMC (p of Stereo D vs. F = 0.14, p of Compressed D vs. F = 0.07, Mono D vs. F = 0.01 and Wide-angle mosaic D vs. F = 0.01).
DME evaluations using all studied digital variants were similar to those from film stereo for HE and RT. However, digital imaging modified by compression, removal of stereo, or substitution of wide-angle mosaics substantially lowered CSME sensitivity and RTMC.
This PDF is available to Subscribers Only