Leveraging Cataract Detection Algorithms to Assess the Effectiveness of Remote & Automatic Cataract Screening Within the Ghanaian Context

Abstract

Cataracts are a condition of the eye where the eye lens becomes cloudy and difficult to see through. They can be managed through early detection and surgery. However, they are the leading cause of blindness globally due to factors such as ignorance and lack of access to screening facilities or technology. This study aims to assess the feasibility and potential effectiveness of remote cataract screening via a smartphone application. We modified a pre-existing cataract detection Convolutional Neural Network (CNN) model to better suit it to the dataset on which we trained it. We integrated the model into a Flutter mobile application which we used to assess user perceptions on the application’s usefulness and overall effectiveness among Ghanaian eyecare specialists and other potential users. We found a generally positive reaction to the concept among both categories of users albeit with a few concerns and suggestions for improvement to better suit the needs of the Ghanaian population. These findings highlight the promising future that mobile-powered cataract screening has in Ghana and its potential to revolutionise eyecare provision, especially in settings with limited resources.