Deep learning based Classification network and Versatile seeded region growing Approach for detection of Age related macular Degeneration in Two Dimensional Retinal Images

Abstract

Abstract Age Related Macular Degeneration (ARMD) is an eye sickness prompts obscured territory close to the focal point of vision or it might create clear spots in the focal vision, and articles may not show up as splendid as they used to be and loss of vision. The WHO revealed that 8 million matured individuals are influenced with ARMD. The infection is essentially grouped into two sorts, wet and dry types of ARMD. The presence of Drusen and Choroidal Neovasculation(CNVs) are the manifestations of dry and wet type of ARMD. The drusen is the yellow-white store related with diminishing or hypo-pigmentation of the retinal pigment epithelium. CNVs lead to draining and scarring causing loss of vision. The diagnosis of ARMD is performed naturally by dissecting the retinal images. In the proposed work, the milestone of retina Optic disc is recognized and veiled by roundabout hough change and morphological methodology. The Drusen is distinguished by nearby and versatile seeded region growing. The sequence of Mathematical morphology operation is utilized to fragment CNVs in STARE database. The performance measures, for example, SE, SPE, and ACC esteems acquired are {98%, 99%, 99%}, {62.86%, 99.39%, 96.95%} for division of drusen and CNVs separately. The deep learning based classification network is used for Classification of Age related Macular degeneration (CO-ARMD)