International Journal of Science, Engineering and TechnologyAuthors: Prasannta Tiwari, Dr. Pritaj Yadav
Abstract: Diabetic Retinopathy (DR) is a progressive eye disease requiring early diagnosis for effective treatment. This study introduces a novel diagnostic framework named GARID (Genetic Algorithm-based Retinopathy Image Diagnosis), designed to enhance the accuracy of retinal image classification through intelligent feature optimization and robust classification. The proposed model operates in two primary phases: feature optimization using a modified Genetic Algorithm (GA) and classification via a Tree Bagger ensemble learning method. Initially, retinal images undergo preprocessing and denoising using Wiener filtering. Segmentation is performed using GA, where cluster centers are evolved through crossover and mutation strategies to identify regions of interest. Features are then extracted using histogram analysis and Discrete Wavelet Transform (DWT), capturing both spatial and frequency information. The final feature set is classified using a tree-based ensemble model, ensuring high generalization and detection precision. Experimental results confirm that GARID improves class-wise detection, recall, and F-measure, offering a reliable solution for automated diabetic retinopathy screening.
