RecursiaDx

AI-powered digital pathology platform for automated tumor detection, malaria detection, and platelet counting with AI-generated clinical reports.

Overview

RecursiaDx is a comprehensive medical image analysis platform that integrates state-of-the-art machine learning models for:

• Tissue Analysis: Tumor detection in histopathology slides using GigaPath-AttentionMIL
• Malaria Detection: Blood smear analysis using InceptionV3
• Platelet Counting: Automated platelet detection using YOLOv11
• AI Report Generation: Gemini AI-powered clinical summary generation

Key Features

✅ Multi-Modal Analysis

• Tissue tumor detection (GigaPath-based AttentionML)
• Malaria parasite detection (Transfer Learning)
• Platelet counting (YOLO object detection)

✅ AI-Powered Workflows

• 5-step clinical workflow (Upload → Analysis → Dashboard → Review → Report)
• Real-time ML inference with interactive visualizations
• Gemini AI-generated clinical summaries and recommendations

✅ Professional Reporting

• Dynamic report generation with AI interpretation
• Morphological findings analysis
• Clinical recommendations
• HIPAA-compliant data handling

✅ Interactive UI

• Dark/Light theme support
• Sample type adaptation (Blood vs. Tissue)
• Real-time status tracking
• Demo mode for testing

Tech Stack

Component	Technology
Frontend	React 18 + Vite + Tailwind CSS
Backend	Node.js 18+ + Express + MongoDB
ML Models	PyTorch + GigaPath + InceptionV3 + YOLOv11
AI Integration	Google Gemini 2.5 Flash
Database	MongoDB Atlas

Quick Start

Prerequisites

• Node.js 18+ and npm
• Python 3.10+
• MongoDB (local or Atlas)
• Google Gemini API key (optional, for AI reports)

Installation

1. Clone repository

   git clone https://github.com/AyushX1602/Recursia-Dx-ML-.git
   cd RecursiaDx

2. Backend setup

   cd backend
   npm install
   cp .env.example .env  # Configure MongoDB URI and Gemini API key

3. Frontend setup

   cd client
   npm install

4. ML setup

   cd ml/api
   pip install -r requirements.txt
   # Download model files (see ml/README.md)

Running the Application

Option 1: Use startup script (Windows)

.\start_all.bat

Option 2: Manual start

# Terminal 1 - Backend
cd backend
node server.js

# Terminal 2 - Frontend
cd client
npm run dev

# Terminal 3 - Tissue ML
cd ml/api
python gigapath_api.py

# Terminal 4 - Blood ML
cd ml/api
python app.py

Access the application at http://localhost:5173

Environment Variables

Backend (.env)

MONGODB_URI=mongodb://localhost:27017/recursiadx
PORT=5001
GEMINI_API_KEY=your_gemini_api_key_here  # Optional
ML_SERVICE_URL=http://localhost:5000
GIGAPATH_SERVICE_URL=http://localhost:5002

ML (.env)

GIGAPATH_MODEL_PATH=path/to/gigapath_model.pth
MALARIA_MODEL_PATH=path/to/InceptionV3_Malaria_PyTorch.pth
PLATELET_MODEL_PATH=path/to/yolo11n.pt

Workflow Steps

1. Sample Upload - Upload tissue/blood images
2. Analysis - ML models process the images
3. Dashboard - View results and visualizations
4. Technician Review - Approve or request re-analysis
5. Report Generation - Generate AI-powered clinical reports

Project Structure

RecursiaDx/
├── backend/          # Node.js API server
│   ├── routes/      # API endpoints
│   ├── models/      # MongoDB schemas
│   ├── services/    # Gemini integration
│   └── server.js    # Entry point
├── client/          # React frontend
│   └── src/
│       ├── components/   # UI components
│       └── lib/         # Utilities
├── ml/              # ML services
│   └── api/
│       ├── app.py            # Malaria/Platelet API
│       └── gigapath_api.py   # Tissue analysis API
└── test/            # Test images

Model Information

Model	Task	Architecture	Accuracy
GigaPath-AttentionMIL	Tissue Tumor Detection	Vision Transformer	~85%
InceptionV3	Malaria Detection	Transfer Learning	~95%
YOLOv11n	Platelet Counting	Object Detection	~90%

API Endpoints

Backend API (Port 5001)

• POST /api/samples/upload - Upload sample images
• POST /api/samples/demo-analysis - Demo mode analysis
• POST /api/reports/generate/:id - Generate report
• POST /api/reports/generate-full/:id - Generate with Gemini

ML APIs

• Tissue: http://localhost:5002/analyze (GigaPath)
• Blood: http://localhost:5000/analyze (Malaria + Platelet)

Gemini Integration

The platform uses Google Gemini 2.5 Flash for:

• Clinical summary generation
• Result interpretation
• Morphological findings description
• Clinical recommendations
• Diagnostic conclusions

Without Gemini API key: System falls back to rule-based summaries.

Documentation

• Startup Guide
• Integration Report
• ML Setup
• Backend API

Contributing

1. Fork the repository
2. Create feature branch (git checkout -b feature/amazing-feature)
3. Commit changes (git commit -m 'Add amazing feature')
4. Push to branch (git push origin feature/amazing-feature)
5. Open Pull Request

License

This project is for educational and research purposes.

Acknowledgments

• GigaPath model by Microsoft Research
• Gemini AI by Google
• Open-source ML communities

---

Status: ✅ Production Ready | 🔄 Active Development

For issues or questions, please open a GitHub issue.