Diabetic Retinopathy Detection Using Capsule Network Modelled Deep Learning

Presently, glaucoma, an acute eye disorder, if untreated, leads to permanent blindness is a major challenging in healthcare system for the complexities of diagnosis. Historical diagnosis models greatly rely on practical assessments of retinal images in addition to, rim width, optic cup and optic disc size, and any abnormalities within the eyes. Advancements of the last few years in AI have brought recent opportunities in diagnosis of glaucoma disorder. The work focuses on a composite learning model that combines UN et++ and CapsN et design for accurate detection and diagnosis of glaucoma. Using UNet++, semantic segmentation is applied together with OD and OC segmented region being critical to the identification of illness. The hybrid model leverage of CapsNet ability to detect hierarchical relationship Successful in better detection of glaucomatous change in comparison to standard Convolutional Neural Networks. The method uses state-of-the-art approaches like CLAHE method to preprocess retinal images, as a result, to improve the image quality. When examined with algorithm models, this research survey has been trained and tested on benchmark datasets for optic curves, optic segmentation and glaucoma detection wherein it has outperformed all of them. This automated solution for glaucoma detection offers promising potential in the fight against blindness, as evidenced by the methods robust trained performance and diagnostic accuracy. The integration in UNet++ and CapsNet Offers a unique and effective solution for prior and accuracy in detection of glaucoma, potentially revolutionizing the field of ophthalmic diagnostics.