Optimized feature selection approach for glaucoma detection using logistic-based chaotic whale optimization algorithm

Abstract

Feature selection in image processing reduces data dimensionality by eliminating unnecessary features. Meta heuristic algorithms inspired by natural phenomena help in accurate classification, mitigating the complexity found in medical diagnosis systems relating to features from a retinal dataset. This research aims to obtain a global optimal solution with a reduced feature set for glaucoma detection using the DRION-DB and DRISHTI-GS1 retinal image datasets and the YiWei Chen Retina Dataset with a decision tree (DT). This study employs a novel optimization approach termed logistic-based chaotic whale optimization algorithm (LCWOA) with an evolutionary search process for the exploration and exploitation phases of chaos theory. An efficient selection algorithm is proposed to improve the classification results of glaucoma images by selecting an optimal feature subset from extracted features using logistic map factor. The proposed method achieves better accuracy than the existing algorithms and state-of-the-art feature selector.