An Automated Detection of Diabetic Retinopathy Using Convolutional Neural Network in ResNet-50

EasyChair Preprint 3002

Abstract

Diabetic retinopathy (DR) is a diabetes complication that affects retinal blood vessels and may lead to blurred vision or even blindness if not diagnosed in early stages. DR is mainly classified as NPDR (Non Proliferative Diabetic Retinopathy) and PDR (Proliferative Diabetic Retinopathy). This problem is occurring in millions of people worldwide. Generally, highly trained clinical experts examine the coloured fundus images to diagnose this disease. This manual diagnosis (by clinicians) is time consuming and error-prone. Therefore, an automated system can be aided to detect diabetic retinopathy quickly for determining the follow-up treatment to prevent blindness. Such automated systems are already developed using the application of machine learning and deep learning algorithms. But these automated systems are not cost efficient and require extensive computational resources. Our proposed work is focused on reducing the computational cost by efficiently using CNN algorithm in ResNet50.

Keyphrases: Convolutional Neural Network, Diabetic Retinopathy, Non Proliferative Diabetic Retinopathy, Proliferative diabetic retinopathy, ResNet-50, deep learning

@booklet{EasyChair:3002,
  author    = {Susmita Mishra and M Manikandan and R Nikhil Raj},
  title     = {An Automated Detection of Diabetic Retinopathy Using Convolutional Neural Network in ResNet-50},
  howpublished = {EasyChair Preprint 3002},
  year      = {EasyChair, 2020}}