The Proposed Convolutional Neural Network Architecture for the Detection and Classification of Eye Diseases
Rahul Singh, Neha Sharma, Rupesh Gupta
Abstract
Early diagnosis and effective treatment of eye diseases require accurate detection and classification. The goal of this research is to create a sophisticated method for identifying and categorizing four primary eye conditions: cataract, diabetic retinopathy, glaucoma, and normal vision. The study employs a novel Convolutional Neural Network (CNN) model developed specifically for this purpose. The dataset used in this study is a large collection of 4217 retinal images divided into four categories. The primary goal is to create a reliable classification model capable of distinguishing between these pathological disorders and normal cases. The CNN architecture has been painstakingly designed to capture the intricate features that are unique to each disease's retinal images. The Adam optimizer, which is well-known for its effectiveness in training deep neural networks, is used to optimize the proposed CNN model. The training protocol has 15 epochs and a batch size of 32, which was chosen for its optimal balance of computational efficiency and convergence to optimal solutions. As a result, the model can extract intricate patterns and features from input images, yielding precise classifications. The experimental results show that the proposed CNN model is capable of detecting and classifying the four types of eye diseases. The model has an impressive 94% accuracy rate, indicating its ability to correctly identify Cataract, Diabetic Retinopathy, Glaucoma, and classify images as Normal. This research advances automated diagnostic tools in the field of ophthalmology, which may help medical professionals make quick and accurate clinical assessments.