Glaucoma Optic Disk Localization and Classification using AlexNet, VGGNet 19, and ResNet 101-based model

Glaucoma is a common eye illness that needs precise localization and categorization of optic disks to provide an accurate diagnosis and appropriate course of therapy. This investigation presents a glaucoma optic disk localization and classification framework that uses deep learning models such as AlexNet, VGGNet 19, and ResNet 101. RIM-ONE v3, ORIGA-Lite, and APTOS 2019 are the three separate datasets used to evaluate the framework. To be ready for the localization task, we enlarge and normalize the RIM-ONE v3 and ORIGA-Lite datasets before isolating the relevant region of the optic disk. We utilize deep learning models to locate the optic disk in the images as accurately as possible. We employ the pre-processed images from the RIM-ONE v3, ORIGA-Lite, and APTOS 2019 datasets to finish the classification task. Images of the optic disk are run through deep learning models like AlexNet, VGGNet 19, and ResNet 101 to determine whether or not the images depict glaucomatous conditions. Glaucoma, diabetic retinopathy, and other eye diseases with varying degrees of severity are all included in the classification procedure. Metrics such as accuracy, sensitivity, specificity, and F1 score are used to assess the suggested framework's level of performance. Several distinct deep-learning models and datasets for glaucoma optic disk localization and classification are subjected to comparative studies to evaluate the efficacy of these models and data sets. The findings highlight the possibility of using deep learning models to properly localize and categorize optic disks for glaucoma diagnosis. The results contribute to the advancement of the area of glaucoma detection and classification by providing insights on the performance and applicability of models such as AlexNet, VGGNet 19, and ResNet 101 concerning datasets such as RIM-ONE v3, ORIGA-Lite, and APTOS 2019.