Automated Detection of Diabetic Retinopathy Using conventional neural network

Abstract

Convolutional neural networks (CNNs) have emerged as a powerful tool in medical image analysis, enabling automated disease detection with high accuracy. In this study, a CNN-based approach was applied to retinal images to detect diabetic retinopathy, a leading cause of vision impairment in diabetic patients. Traditional detection methods rely on manually defined image features, such as blood vessels or exudates, which can limit diagnostic accuracy and require significant human intervention. These approaches also face challenges in identifying subtle pathological variations due to the complex and diverse visual patterns in retinal images.

The proposed CNN model automatically extracts relevant visual features and classifies retinal images without manual intervention, achieving robust performance in distinguishing between Moderate and No Diabetic Retinopathy cases. Furthermore, the system is designed for potential integration into Internet of Things (IoT) environments, allowing real-time, remote diagnostics and supporting improved healthcare delivery. These results demonstrate the potential of CNNs to enhance automated screening and contribute to more efficient, accurate diabetic retinopathy detection.