Vision AI: Enhancing Ophthalmic Care with Advanced Detection Technologies

This study presents Vision AI, an advanced diagnostic framework that utilizes hybrid deep learning models for accurate detection of eye diseases, specifically Glaucoma, Macular Hole, Central Serous Retinopathy, and Drusen. Traditional ophthalmic diagnostics often suffer from human error and inefficiencies, emphasizing the need for automated solutions. Vision AI addresses these challenges by integrating multiple model architectures, such as VGG16, ResNet, V-Net, and YOLOv8x, optimizing diagnostic accuracy and generalizability. A carefully curated dataset of retinal images was used, supported by extensive data augmentation and hyperparameter tuning to enhance model robustness. This approach significantly improved detection capabilities, achieving over 95% specificity and sensitivity across diseases. Additionally, Vision AI includes a Large Language Model for providing tailored health recommendations, making the system highly practical for clinical use. The framework is deployed in a scalable microservice-based web application, ensuring modularity, security, and real-time diagnostics, thus improving accessibility for healthcare providers. Vision AI showcases the potential of hybrid deep learning systems to transform ophthalmic care, laying a foundation for AI-assisted, patient-centered healthcare advancements.