Purchase this article with an account.
C.J. Rudnisky, M.D. Greve, P. Sanghera, M.T. Tennant, N. Heyns, B.J. Hinz, A.R. DeLeon; Multifield Non-Mydriatic Digital Fundus Photography Versus High-Resolution Stereoscopic Digital Imaging for the Tele-Ophthalmologic Diagnosis of Diabetic Retinopathy . Invest. Ophthalmol. Vis. Sci. 2003;44(13):4283.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: To compare nonmydriatic digital retinal photography, the method used by most tele-ophthalmology groups, with high-resolution stereoscopic digital imaging for the diagnosis and management of diabetic retinopathy. Methods: Following informed consent and prior to pupillary dilation, 36 patients (72 eyes) from a remote northern Alberta community had three digital nonmydriatic fundus photographs captured per eye. Following mydriasis, seven 30 degree high resolution digital images were obtained, with Airlee House Fields 1 and 2 captured in stereo. Both formats were archived, transported and then randomly reviewed in Edmonton for image quality, microaneurysms, intraretinal hemorrhage, hard exudates, cotton-wool spots, intraretinal microvascular abnormalities, venous beading, clinically significant macular edema (in the high-resolution stereo images), presumed CSME (in the non-stereo nonmydriatic images), neovascularization, and vitreous and preretinal hemorrhage. The level of diabetic retinopathy and recommended clinical follow-up was also recorded. To limit reader bias, the reading of the nonmydriatic images and high-resolution images was conducted a minimum of 2 weeks apart. Results: The nonmydriatic camera had good or excellent correlation (kappa statistic 0.484 - 0.811) with high-resolution stereoscopic digital imaging for all pathologies except CSME and microaneurysms. Similarly, the nonmydriatic camera had good or better correlation for all diagnoses in this cohort (kappa statistic 0.424 - .0525) except very mild NPDR. The sensitivity of the nonmydriatic camera ranged from 25% (CSME) to 75% (intraretinal hemorrhage). Specificity was high for all measured pathologies ranging from 91.5% to 100%. However, the nonmydriatic camera produced 5 times as many ungradeable images and required clinical re-examination for nearly three times as many patients. Conclusions: Because of both better image quality and stereopsis, high-resolution stereoscopic digital imaging detects more treatable diabetic retinopathy and results in fewer rural diabetic patients requiring needless referral for clinical re-examination. High-resolution stereoscopic digital imaging may result in a cost benefit for health care providers.
This PDF is available to Subscribers Only