Traditional ophthalmic imaging techniques, such as fundus fluorescein angiograms and indocyanine green angiography, are crucial for diagnosing retinal conditions but are invasive and uncomfortable for patients. This project aims to develop and clinically validate advanced AI-based software as a medical device for non-invasive ophthalmic imaging and automated diagnostic reporting. The study applies generative adversarial networks (GANs) to transform standard fundus photographs into angiography-like images, eliminating the need for invasive procedures. The work focuses on refining these AI models, integrating them into a user-friendly and regulatory-compliant platform, and conducting rigorous clinical validation for retinal conditions such as diabetic retinopathy and age-related macular degeneration.