Purchase this article with an account.
Samuel Stratton, Jorge Luna, Shiyoung Roh, Noreen Shaikh, Amer Alwreikat, Ying Jiang, Paul Cotran, Mahesh Bhardwaj, David J Ramsey; Smartphone-based Fundus Photography for Remote Glaucoma Assessment in a Low-Resource Setting. Invest. Ophthalmol. Vis. Sci. 2021;62(8):1616.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
This investigation examines the effectiveness of smartphone-based, nonmydriatic fundus photography in detecting glaucoma in patients in a low-resource setting and explores the potential for this technology to serve as a means of primary screening for risk of glaucoma.
Subjects between the ages of 40 and 70 years old were recruited to undergo nonmydriatic fundus photography during a visual health campaign at the Salvadoran Association for the Promotion of Rural Health (ASAPROSAR) clinic in Santa Ana, El Salvador in January 2020. Color fundus photographs were obtained with a PanOptic iExaminer attached to an iPhone 6S operated by a trained non-specialist. Stereoscopic image pairs were assembled from two “best view” images for each eye. These were graded by a team of ophthalmologists and optometrists experienced in teleophthalmology who were blinded to the clinical diagnosis of each patient.
A total of 273 participants were screened. Gradable images were obtained from 509 eyes (93.2%) in 268 participants (98.1%). A total of 5,744 images were collected, averaging 10.5 ± 5.2 images per eye. The cup-to-disc ratio (CDR) judged from the stereoscopic images showed moderate reliability and good agreement with the CDR reported by the treating ophthalmologists (ICC 0.59; r = 0.792, p < 0.001). Overall agreement was fair for glaucoma risk determined by remote grading of stereoscopic image pairs compared with the glaucoma-related diagnoses by ophthalmologists who examined the participants (kappa = 0.32 ± 0.09). Accuracy of screening for risk of glaucoma using only features of the optic disc observable in stereoscopic images was low (66.8%), with a positive predictive value of 45.1% and negative predictive value of 87.7%, compared with the results from clinical examination.
Remote evaluation by experts of CDR from optic disc photographs obtained by using a low-cost, handheld non-mydriatic fundus camera operated by a non-specialist showed good agreement with CDR assessment from in-person clinical examinations. As expected from previous studies, the performance of CDR alone in detecting glaucoma was only fair. Nevertheless, the capacity to obtain high-quality fundus photographs and generate stereoscopic image pairs for remote evaluation by expert graders could permit development of an affordable, high-volume, glaucoma screening strategy in resource-constrained areas of the world.
This is a 2021 ARVO Annual Meeting abstract.
This PDF is available to Subscribers Only