INTELLIGENT SYSTEM FOR DIAGNOSIS OF OPHTHALMOLOGICAL DISEASES

Abstract

The article presents an innovative intelligent system for the automated diagnosis of ophthalmic diseases based on the analysis of fundus images using machine learning and deep learning methods. The aim of the work is to develop a platform that integrates medical devices, cloud technologies, and artificial intelligence algorithms to improve screening and early detection of diseases such as age-related macular degeneration (AMD), diabetic retinopathy (DR), glaucoma, and pathological myopia (PM). 
The relevance of the research is due to the growing prevalence of ophthalmic pathologies, which often lead to irreversible vision impairment as a result of untimely diagnosis. Existing systems typically operate autonomously and are not integrated with medical devices, which limits their applicability. The proposed platform addresses this issue by combining the capabilities of cloud computing, modern image processing algorithms (convolutional neural networks – CNNs, support vector machines – SVM), and a user-friendly interface for both patients and doctors. 
The methodology includes the following stages.Data upload: Patients upload retinal images and medical data via a web portal or mobile application. Preprocessing: The images undergo segmentation, normalization, and feature extraction using CNNs. Classification: The SVM algorithm analyzes the extracted features to determine the presence and type of disease. Report generation: The system generates a diagnostic report with recommendations, available to both the patient and the doctor. 
The system architecture includes the following component. Input layer: an interface for data upload. Processing layer: a convolutional neural network for feature extraction (convolutional layers, ReLU, pooling) and a support vector machine for classification. Output layer: A web portal with analytical reports. 
The results confirm the system's effectiveness: it provides fast and accurate analysis with minimal prediction time, which is especially important for regions with a shortage of ophthalmologists. The model was trained on a dataset covering all target pathologies, ensuring high diagnostic specificity.