DynamicXRayCT: Tomographic Reconstruction of Discrete Dynamic Objects

Ajinkya Kadu · Felix Lucka · Joost Batenburg

IEEE TCI

Welcome to DynamicXRayCT! We introduce a novel method that integrates level-set method and compressed sensing for the reconstruction of temporal images in dynamic tomographic imaging.

🚀 Dynamic Tomographic Reconstruction

In dynamic tomographic imaging, capturing discrete objects with smooth boundaries that vary over time has always been a challenging task, especially with limited measurements at each time point. DynamicXRayCT addresses this challenge by advanced algorithm. Our approach combines spatial and temporal information of dynamic objects, using the level-set method for image segmentation and a sinusoidal basis for motion representation. This results in a computationally efficient and easily optimizable variational framework called Dynamic Shape Sensing (DSS). Our method stands out by enabling the reconstruction of high-quality 2D or 3D image sequences with just a single projection per frame.

🌈 Key Features

• Temporal Image Reconstruction: Achieve high-resolution imaging with limited measurements.
• Level-Set Image Segmentation: Segmentation for smoothly evolving objects.
• Sinusoidal Motion Capture: Captures the dynamics of objects in motion.
• Optimized Variational Framework: Offers computational efficiency and ease of optimization.
• Xray Dataset Compatibility: Demonstrate better performance on both synthetic and real X-ray tomography datasets.

Please consider citing our work if you find it useful

@article{Kadu:DynamicTomo:2023,
  title={Single-shot Tomography of Discrete Dynamic Objects},
  author={Kadu, Ajinkya and Lucka, Felix and Batenburg, Kees Joost},
  journal={arXiv preprint arXiv:2311.05269},
  year={2023}
}

🛠 Getting Started

Please refer to the following for setting up and running the code:

• Installation Instructions
• Usage Guidelines

Repo Structure:

• 📁 /src: the source codes.
• 📁 /examples: CT applications and examples.
• 📁 /docs: documentation and resources.

📚 Citation

@article{Kadu:DynamicTomo:2023,
  title={Single-shot Tomography of Discrete Dynamic Objects},
  author={Kadu, Ajinkya and Lucka, Felix and Batenburg, Kees Joost},
  journal={arXiv preprint arXiv:2311.05269},
  year={2023}
}

©️ License

The code and models are available for use without many restrictions. See the LICENSE file for details.

💡 Contributions and Feedback

Please contact Ajinkya Kadu for any questions. We welcome contributions and feedback from the community. We invite you to:

• 🐛 Report Issues
• 🌟 Suggest Enhancements
• 🤝 Contribute to the Code

Check our Contribution Guidelines for more.