Smartphone funduscopy for diabetic retinopathy screening

Abstract

Aims: Smartphone funduscopy devices are comprised of lens devices connecting with smartphones and software applications to be used for mobile retinal image capturing and diagnosis of diabetic retinopathy. This is particularly beneficial to automate and mobilize retinopathy screening techniques and methods in remote and rural areas as those diabetes patients are often not meeting the required regular screening for diabetic retinopathy. Methods: A novel smartphone fundoscope was developed as a prototype. The lens is attached into a sterile plastic tube which securely holds the lens at the correct fixed distance from the smartphone. The tube is attached onto a smartphone case which enables it to fit tightly onto the phone. The camera is set to record video, and LED flash torch is switched on. A second prototype has a screw cap that can extend or retract the lens for fine tuning the image focus. Results: The prototype 1 was tested and produces suitably high resolution microscopic zoom images. Conclusions: A smartphone app prototype was developed that can obtain fundus images by interfacing with the developed lens attachment. This app could be extended in future to apply a pre-trained machine learning model (neural network) within the smartphone to give stand-alone diagnosis. If the smartphone app is unable to classify the fundus image, images can be uploaded onto a cloud based classification system, and to forward the fundus image to the ophthalmology expert for manual viewing and diagnosis.