Enhance Diabetic Retinopathy Prediction based on Retinal Fundus Image using Machine Learning Techniques

Abstract

Diabetic Retinopathy (DR) is a severe eye condition that can lead to blindness if left undiagnosed and untreated. With the increasing number of individuals affected by diabetes, scaling up DR screening has become a challenging task. This research aims to leverage deep learning techniques, specifically Convolutional Neural Networks (CNNs), to address the challenges associated with DR detection and classification. However, traditional CNN models often face limitations in categorizing DR images due to overfitting, leading to poor performance. In this study, we propose a novel approach that combines the strengths of CNN and adaptive boosting algorithms to overcome the overfitting problem and achieve improved accuracy. The proposed model leverages a pre-trained VGG16 model to learn robust features from DR images, which are then preprocessed using a factor analysis method. Subsequently, a CNN-based adaptive boosting approach is focused on an accurate categorization of the learned features. newlineThis research aims to investigate the efficacy of deep learning techniques at different stages of DR diagnosis using fundus images. The proposed hybrid DRNN (Diabetic Retinopathy Neural Network) method focuses on detecting early-stage diabetic retinal degeneration. Experimental results demonstrate that our suggested model achieves an impressive training accuracy of 99.86% and a testing accuracy of 96.91%. Additionally, we compare the performance of the proposed DRNN model with other deep learning methods such as CNNs, RNNs, DBNs, and Google Nets. The comparative analysis showcases the superior performance of our DRNN model and provides valuable insights for future algorithm selection in retinopathy detection and classification research. newline newline newline