Internet of Things and Deep Learning Enabled Diabetic Retinopathy Diagnosis Using Retinal Fundus Images

The proposed system aims to enhance the early detection and classification of diabetic retinopathy and macular edema using advanced image processing and deep learning techniques. The process begins with the input of Fundus images, which undergo preprocessing steps such as resizing, noise removal, and contrast enhancement to prepare the data for analysis. A Mayfly Optimization-based Region Growing (MFORG) algorithm is employed to identify key regions within the image. Following this, a Dense Convolutional Network (DenseNet) is utilized for feature extraction, followed by classification into healthy or diabetic retinopathy-affected retina. If diabetic retinopathy is detected, the system further classifies the condition into four stages: Mild, Moderate, Severe, and Proliferative Diabetic Retinopathy (PDR). For macular edema classification, the system applies morphological closing, adaptive contrast enhancement, Gabor kernel-based filtering, and thresholding using the OTSU algorithm to create a binary map. This map is processed by a Convolutional Neural Network to classify the macular region as either normal or abnormal. Based on these classifications, the system generates tailored treatment options and lifestyle recommendations, which are automatically sent to the patient via email. Additionally, the system logs data into a ThingSpeak channel for real-time monitoring, visualization, and tracking of the patient's condition, facilitating ongoing analysis and adjustments to treatment plans as necessary. This comprehensive approach not only aids in accurate diagnosis but also supports effective patient management through continuous data monitoring and personalized treatment recommendations.

Keywords: Fungus Dataset, Deep Learning, Convolution Neural Network, Image Processing Techniques, segmentation and accuracy.