Healthway Diagnostics - Neural Retina Analysis

Live Demo: https://retina-25s2.vercel.app

Healthway Diagnostics is a professional-grade clinical platform that leverages deep learning to detect Diabetic Retinopathy (DR) from fundus photography (retina scans). Using an EfficientNet-B0 neural engine, it provides instant diagnostic classification and visual "heatmaps" (Grad-CAM) to assist medical professionals.

---

🏗️ Architecture

The system operates on a three-tier architecture:

1. AI Engine (Python/FastAPI): High-performance inference server using PyTorch.
2. Clinical Gateway (Node.js/Express): Middleware handling patient records, file storage (Multer), and PDF report generation (PDFKit).
3. Diagnostic Portal (Next.js/TypeScript): A modern, clinical-grade interface with real-time analysis status and history tracking.

---

✨ Key Features

• Neural Diagnosis: Classifies scans into 5 stages: Healthy, Mild, Moderate, Severe, or Proliferative DR.
• Explainable AI (XAI): Generates Grad-CAM heatmaps to highlight pathological markers (Hemorrhages, Exudates, etc.).
• Clinical Reports: Generates professional PDF diagnostic reports for patients.
• History Management: Securely stores and retrieves past patient analyses using MongoDB.
• Image Validation: Automated biological texture checking to prevent non-retinal uploads.

---

🛠️ Tech Stack

• Frontend: Next.js 15, React 19, TypeScript, Tailwind CSS, SweetAlert2.
• Middleware: Node.js, Express, MongoDB (Mongoose), PDFKit.
• AI Backend: Python 3.10+, FastAPI, PyTorch, Torchvision, OpenCV.

---

🚀 Local Setup

1. Neural Engine (AI Service)

cd backend
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt
python main.py

Runs on http://127.0.0.1:8080

2. Clinical Gateway (Middleware)

cd server
npm install
# Create a .env file with:
# MONGO_URI=your_mongodb_uri
# AI_SERVICE_URL=http://127.0.0.1:8080
node index.js

Runs on http://127.0.0.1:5000

3. Diagnostic Portal (Frontend)

npm install
# Create a .env.local file with:
# NEXT_PUBLIC_API_URL=http://127.0.0.1:5000
npm run dev

Runs on http://127.0.0.1:3000

---

🌐 Deployment

This project is pre-configured for deployment on Vercel (Frontend) and Render/Railway (Backend & Middleware).

1. Frontend: Deploy the root directory to Vercel. Set NEXT_PUBLIC_API_URL.
2. Middleware: Deploy the /server folder. Set MONGO_URI and AI_SERVICE_URL.
3. AI Service: Deploy the /backend folder. No extra config required.

---

⚠️ Disclaimer

This software is for educational/research purposes only. It is not a certified medical device. The diagnostic results provided by the neural engine should always be reviewed by a certified ophthalmologist or medical professional.

---

📄 License

Distributed under the ISC License. See LICENSE for more information.