README.md

metadata

title: SimEIT Datasets Visualizer
emoji: 🔬
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 5.49.1
app_file: app.py
pinned: false
license: apache-2.0
tags:
  - medical-imaging
  - electrical-impedance-tomography
  - eit
  - dataset-visualization
  - simulation
  - pytorch
  - computer-vision
short_description: Visualizer for SimEIT synthetic EIT datasets

SimEIT: Large-Scale Electrical Impedance Tomography Dataset Visualizer

A Scalable Simulation Framework for Generating Physically Consistent, AI-Ready EIT Training Data

Ayman A. Ameen¹, Franziska Mathis-Ullrich¹, Bernhard Kainz²

¹Friedrich-Alexander University Erlangen-Nürnberg
²Imperial College London

---

🎯 About This Demo

This interactive demo allows you to explore large-scale synthetic EIT (Electrical Impedance Tomography) datasets generated using the SimEIT framework—a scalable simulation platform for creating physically consistent, AI-ready training data.

🔬 What is SimEIT?

Electrical Impedance Tomography (EIT) offers advantages over conventional imaging methods, such as X-ray and MRI, but suffers from an ill-posed inverse problem. Deep learning can alleviate this challenge, yet progress is limited by the lack of large, diverse, and reproducible datasets.

SimEIT enables high-throughput creation of diverse geometries and conductivity maps using parallelized finite element simulations, reproducible seeding, and automated validation. The framework provides multi-resolution, AI-ready HDF5 outputs with PyTorch integration, bridging the gap between physical simulation and AI training.

✨ Features

• 🔄 Streaming Mode: Load datasets directly from Hugging Face Hub without downloading
• 🖼️ Multi-resolution Images: View images at different resolutions (256×256, 128×128, 64×64, 32×32)
• 📊 Interactive Voltage Plots: Visualize voltage data per electrode with Plotly
• 🎨 Customizable Colormaps: Choose from 18 different scientific colormaps
• 🎲 Flexible Selection: Choose samples randomly or by specific index
• 💾 Efficient Caching: LRU cache for fast repeated access to samples
• 📈 Two Dataset Variants: Explore 'FourObjects' or 'CirclesOnly' subsets

🚀 How to Use

1. Select Dataset Configuration: Choose between 'FourObjects' or 'CirclesOnly' subsets
2. Choose a Sample:
   • Click "Generate Random Index" for a random sample
   • Or enter a specific index (0-100,000)
3. Customize Visualization:
   • Select your preferred image resolution
   • Choose a colormap for visualization
   • Toggle between linear and log scales
4. View Results:
   • Explore multi-resolution conductivity and permittivity maps
   • Analyze electrode voltage measurements
   • Examine the domain geometry

📊 Dataset Information

The SimEIT dataset contains:

• 100,000+ samples per subset
• Multi-resolution images: 256×256, 128×128, 64×64, 32×32
• Physical parameters: Conductivity, permittivity, electrode voltages
• Geometry information: Object positions and boundaries
• Two subsets:
  • FourObjects: Complex scenes with up to 4 objects
  • CirclesOnly: Simplified circular objects

🔗 Links

• 📦 Dataset: AymanAmeen/SimEIT-dataset
• 💻 Code Repository: GitHub
• 📄 Paper: ayman-ameen.github.io/SimEIT_page

🛠️ Technical Details

This visualizer uses:

• Gradio for the interactive interface
• HDF5 for efficient data storage and streaming
• Plotly for interactive plots
• Hugging Face Hub for seamless dataset access
• NumPy and OpenCV for image processing

📝 Citation

If you use the SimEIT dataset or framework in your research, please cite:

@misc{simeit2025,
  title={SimEIT: A Scalable Simulation Framework for Generating Physically Consistent, AI-Ready EIT Training Data},
  author={Ameen, Ayman A. and Mathis-Ullrich, Franziska and Kainz, Bernhard},
  year={2025},
  institution={Friedrich-Alexander University Erlangen-Nürnberg, Imperial College London}
}

📧 Contact

For questions or feedback:

• Ayman A. Ameen: just drop an email.
• Issues: Please open an issue on GitHub

📄 License

This project is licensed under the apache-2.0 License. See the LICENSE file for details.

---

Built with ❤️ for the research community