Abstract
Purpose :
This investigation examined the effectiveness of smartphone-based, nonmydriatic fundus photography in detecting diabetic retinopathy in patients in a low-resource setting.
Methods :
Patients were recruited to undergo nonmydriatic fundus photography during a visual health campaign in rural El Salvador in January 2020. After written informed consent, color fundus photographs were obtained by using the PanOptic iExaminer attached to an iPhone 6S (Apple Computer, Inc., Cupertino, CA) operated by a trained, lay examiner. Two high-quality, suitably aligned views of the fundus were selected to be reconstructed into stereoscopic image pairs. These image sets were viewed in stereo by using a stereo viewer (Pocket Stereo Viewer, Luminos Photo Corp., Yonkers, NY) placed over the iPhone 6S screen. Stereoscopic image pairs were graded by a retinal specialist experienced in diabetic screening remote to the clinic and blinded to the diagnosis for each patient.
Results :
Out of 273 participants who underwent screening by means of nonmydriatic fundus photography, 48 participants were identified to have diabetes (18%). Ten of those participants (21%) were noted to have more than mild diabetic retinopathy (mtmDR) based upon a comprehensive eye examination. The sensitivity of smartphone-based nonmydriatic fundus imaging to detect mtmDR was 77.8% with a specificity of 94.7%. The overall accuracy of screening for diabetic retinopathy based only upon features observable in nonmydriatic, stereoscopic image pairs was moderate (89.3%).
Conclusions :
Fundus images were consistently obtained using a portable, low-cost nonmydriatic camera attached to a smartphone in an underserved, rural population of patients with diabetes. While smartphone-based nonmydriatic fundus imaging displayed efficacy, the overall accuracy for diabetic retinopathy screening, based solely on observable features in stereoscopic image pairs, was modest. Our findings underscore the importance of following up screening with comprehensive eye examinations, whenever possible, to ensure a thorough assessment of diabetic retinopathy.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.