Image enhancement optimization on bright and dark spots of retinal fundus image

Abstract

Diagnosing diabetic retinopathy (DR) based on features that appear on fundus images is currently conducted through an eye exam by an ophthalmologist. Tracking DR progression manually is time-consuming and keen for a high-skill person. As the technology offered in industrial revolution (IR) 4.0, namely artificial intelligence, is shown to help in the medical diagnosis process, this study proposes an image enhancement algorithm based on a hybrid of contrast enhancement (CE) and particle swarm optimization (PSO). The proposed method incorporates contrast adjustment on the bright and dark region of LAB color space where the bright and dark region is initially segmented using K-mean PSO. 100 retinal fundus images are used for training and testing purposes. The proposed method undergoes qualitative and quantitative evaluation with a comparison between the two methods. The result indicates that the performance of the proposed method is more acceptable as compared to another two methods.