Retinal Image-Based Deep Learning Neural Network Approaches for Glaucoma Detection: Analysis on Improved Accuracy and Processing Power

Glaucoma, an ocular disorder, has the potential to inflict destruction upon the optic nerve. The optic nerve facilitates the transmission of visual information form the brain to the eyes. The phenomenon known as intraocular pressure (IOP) refers to the accumulation of pressure within the eye, which can ultimately result in harm to the optic nerve. Failing to recognize glaucoma in its earliest phases can potentially result in the onset of blindness or visual impairment. The utilization of fundus pictures is utilized within the suggested study to facilitate the automation of glaucoma diagnosis. The article discusses the development of a hybrid model combining the VGG16 (Visual Geometry Group) and convolutional neural net-work (CNN) architectures for the purpose of glaucoma detection. The algorithm is educated through its utilization of fundus photographs, and subsequent to its instruction, it is capable of discerning among eyes that are healthy and those af-flicted with glaucoma by analyzing the features extracted from the input photos. An aggregate of 520 photos were utilized in order to enhance the precision and computational capabilities of the deep learning approach. The assessment of the CNN and VGG16 models is conducted by assessing their efficacy in terms of precision, robustness, exactness, and F1 score. The Glaucoma detection approach is frequently utilized in the analysis of color fundus pictures. The CNN and VGG16 models achieved a success rate of $76 \%$, a loss of 4.83, a sensitivity of $78 \%$, a precision of $93 \%$, and an F1 score of 1.03. The findings demonstrate the efficacy of the suggested technique for quick execution in realworld scenarios.