Integrated Deep Learning Methodology for Early Glaucoma Detection and Diagnosis using Retinal Fundus Images

This study proposes a thorough methodology for diagnosing glaucoma in retinal fundus pictures using modern image processing and deep learning techniques. The photos are resized using Nearest Neighbor Interpolation before being converted to grayscale, noise reduced with a Kalman Filter, and contrast enhanced with Dualistic Sub-Image Histogram Equalization. Scale-Invariant Feature Transform (SIFT) is used to extract features, which are then processed by VGS Net, a deep learning network developed exclusively for glaucoma diagnosis. The methodology has been confirmed by extensive performance analysis, with an accuracy of 99.06%, precision of 99.19%, recall of 98.13%, and an F1 score of 97.65%. These findings show that the proposed technique is both effective and reliable, making it a promising tool for detecting early glaucoma in clinical settings.