Review Paper on Glaucoma Detection Using Machine Learning

Glaucoma is one of the leading causes of vision loss worldwide. Glaucoma cannot be cured in its advance stages. So, early detection of disease has become an important factor in the medical field. Numerous studies quickly became clear that using different image processing methods, the retinal fundus picture could be uncovered. In this study, many automated glaucoma detection techniques were thoroughly reviewed various papers were compared on the basis of the methodologies they adopted for detecting glaucoma from 2D fundus images created using CDR. 85% of glaucoma cases can be accurately detected by the majority of machine learning algorithms. First, image segmentation techniques like Elliptical Hough transform and edge detection gave the region of interest, i.e. optic disc and cup. These extracted images were then given to the machine learning and deep learning models to detect presence of glaucoma in the fundus image of the eye. The most significant deep learning, machine learning, and transfer learning methods for analysing retinal images were reviewed, along with their benefits and drawbacks.