Optic Pro: AI-Powered Retinal Disease Detection System

Abstract

OptiPro represents a comprehensive web based explainable artificial intelligence platform de signed for automated retinal disease analysis and enhanced clinician workflow support. Our system integrates a ResNet 101 convolutional neural network alongside Gradient weighted Class Activation Mapping technology to accurately classify fundus images into four distinct cat egories: CNV, DME, DRUSEN, and NORMAL. More importantly, the system provides visual rationales that clearly highlight decision relevant retinal regions, addressing a critical barrier to clinical adoption by allowing medical professionals to verify and trust model outputs. The platform architecture follows a three tier implementation approach. The backend utilizes a Flask API built with Python, PyTorch, and OpenCV to handle image preprocessing, model inference, and Grad CAM heatmap generation. The frontend consists of a React based single page application that provides doctors with an intuitive panel for patient management, image uploads, results review, and comprehensive reporting. Data persistence and security are managed through Supabase, which provides PostgreSQL database services and cloud storage capabilities. This work contributes a complete end to end production ready solution that successfully com bines accurate computer vision capabilities with transparent explanations and a clinician cen tered workflow design. The system design prioritizes patient privacy through row level security implementations, comprehensive auditability features, and practical operational considerations including automated report generation, configurable doctor availability management, and intel ligent conflict checking. Our research demonstrates how explainable artificial intelligence can be effectively integrated into routine ophthalmic practice, significantly improving diagnostic throughput and consistency while maintaining essential clinician oversight. Potential future en hancements include expanding condition coverage, integrating optical coherence tomography data, and conducting prospective clinical validation studies.