Glaucoma Disease Detection Using Hybrid Deep Learning Model
J. Manikandan, Sidharth Raj M, R Yogeshkumar, S C
Abstract
Glaucoma is a chronic, irreversible eye disease that affects vision and diminishes the quality of life. The goal of this initiative is to identify the illness sooner so that the person does not completely lose their vision. We cannot restore vision that has been lost due to glaucoma. The current study focuses on constructing a Hybrid Deep Learning Algorithm called CAPSGAN that functions as an effective glaucoma detection tool to create a toolkit for diagnosing the disease. The glaucoma dataset from the Kaggle repository was used in the current study work for accurate detection. There were two key components to the proposed CAPSGAN. In the first step, synthetic images are created using the Generative Adversarial Network (GAN), which primarily aims to provide more image data that will be used for the classification process. The Caps-Net has shown to be the best option for effective picture categorization, outperforming the well-known CNN and a few other machine-learning models. In contrast to simple CNN, which tends to lose important information during the max pooling phase, Caps-Net is specifically built to preserve the spatial, locational, and orientational specifics of image data, which are crucial for accurately determining whether or not a person is infected with the disease. A modified version of the Capsule Network (CAPSNET) is used for classification.