AntarEye
Status: Completed
AntarEye—focused on enabling faster and more accessible eye disease screening for small clinics. The system: • Takes a fundus image as input • Predicts probabilities for: • Diabetic Retinopathy (with severity grading) • Glaucoma • Hypertensive Retinopathy • Generates: - A patient-friendly report (simple, actionable language) - A doctor report with technical insights, heatmaps, and suggested next steps
Technologies
Full Stack, Next.js, TypeScript, FastAPI, PostgreSQL, PyTorch (ResNET18), Google Gemini API, Supabase, Cloudinary
Key Features
- Secure doctor authentication and patient profile management handling.
- Upload fundus retinal images for asynchronous AI inference.
- Detect Diabetic Retinopathy severity from None to Proliferative.
- Assess binary condition risks for Glaucoma and Hypertensive Retinopathy.
- Generate Grad-CAM heatmaps to highlight visual regions of concern.
- Create LLM-powered structured clinical interpretations for general practitioners.
- Generate simplified, patient-friendly diagnostic reports for easy understanding.
- Store and download both clinical and patient reports as PDFs.
- Intelligent triage queue to surface high-priority cases for review.
Tags
Hackathon, Full Stack, Healthcare