Detection and Classification of Neovascularization in Proliferative Diabetic Retinopathy Using Deep Learning

Main Article Content

Pankaj Kumar
Lal Singh

Abstract

Proliferative Diabetic Retinopathy (PDR) is a severe retinal disorder that poses a substantial risk to diabetic patients. Among the various complications associated with PDR, neovascularization stands out as a critical condition characterized by the abnormal growth of blood vessels on the retina. If left undetected and untreated in its early stages, neovascularization can lead to severe vision loss and even blindness. This paper explores the potential of utilizing fundus images, which provide a detailed view of the eye's posterior, for the early detection and classification of neovascularization. Given the intricate and random growth pattern and small size of neovascularization, conventional image processing techniques face challenges in accurate detection. In response to the inherent challenges associated with the precise detection and classification of neovascularization in fundus images, this study endeavors to harness the capabilities of deep learning methodologies. Deep learning has demonstrated exceptional proficiency in the automated extraction of pertinent features from complex objects characterized by intricate attributes. Specifically, the present investigation introduces a novel system, leveraging the prowess of pretrained deep neural networks that have garnered prominence within the field. The ensemble of pre-eminent neural architectures encompasses Inception ResNetV2, DenseNet, ResNet50, ResNet18, AlexNet, and VGG19, collectively representing a comprehensive spectrum of deep learning models.

Article Details

Section
Articles