DEVELOPMENT AND ANALYSIS OF DEEP LEARNING MODEL BASED ON MULTICLASS CLASSIFICATION OF RETINAL IMAGE FOR EARLY DETECTION OF DIABETIC RETINOPATHY

Abstract

Diabetic retinopathy (DR) is a leading cause of blindness, and early detection is crucial for effectively managing and preventing vision loss. This paper proposes a deep learning-based model for the early detection of diabetic retinopathy (DR) using retinal images. The proposed model uses a convolutional neural network (CNN) architecture and transfer learning-based EfficientNet architecture for multiclass classification (0- No DR, 1- Low, 2- Medium, 3- High, 4- Proliferative) of DR, on a large dataset of annotated retinal images. The performance of the model is evaluated on an independent test set and compared with CNN and EfficientNet methods. Results show that the efficient model achieves high accuracy and outperforms existing methods for DR detection. Moreover, the model can detect DR at an early stage, enabling timely interventions and preventing vision loss. The results show that we achieved a training accuracy of 94.42% after 20 epochs and a testing accuracy of 81.81%. The model's accuracy and early detection capability make it a promising tool for enhancing DR screening programs and enabling timely interventions to prevent vision loss.