X-Ray Path Tracing for CT Imaging

Bachelor Thesis
Student Project
Master Thesis
1-2

Description

The tracing of light paths through 3D scenes is the de-facto standard for generating photorealistic images – both in research and industry. By specifying the interaction of light with surface materials and volumetric media, Monte Carlo-based methods allow the accurate simulation of light transport even in challenging cases such as caustics and participating media. Moreover, spectral and polarization effects can be handled as well.

This project aims to transfer these capabilities, which were originally developed for visible light, to the X-ray spectrum. By specifying the spectral attenuation of X-ray inside volumetric media and by considering metal surfaces, advanced effects – such as beam hardening and Compton scattering – should be achievable with path tracing. This project will have immediate practical applications as it can be used to simulate 3D X-Ray Computed Tomography (XCT) – a well-established and wide-spread imaging method in medicine and industry. By extending a state-of-the-art path tracer, we expect significant performance gains over existing XCT simulation software [1].

After having achieved a realistic XCT simulation, the second step of this project is to utilize the recently achieved differentiability of path tracing to inverse the X-Ray rendering pipeline [2]. This ambitious step – which requires the time frame of a diploma thesis – holds the potential for groundbreaking advances in the reconstruction quality of XCT images. This has immediate medical and industrial application and would pave the way for interesting follow-up work.

Requirements

  • Solid understanding of Monte Carlo rendering
  • Interest in learning X-Ray physics
  • Good programming skills in C++ and Python
  • Sufficient time for this project

Perks:

  • Work on cutting-edge rendering software [3]
  • Learn cool physics from experts [4]
  • Develop a highly relevant application
  • Potential for future continuation

References:

  1. SimCT: http://www.3dct.at/cms2/index.php/en/software-en/simct
  2. Merlin Nimier-David, Delio Vicini, Tizian Zeltner, and Wenzel Jakob. 2019. Mitsuba 2: a retargetable forward and inverse renderer. ACM Trans. Graph. 38, 6, Article 203.
    DOI: https://doi.org/10.1145/3355089.3356498
  3. Mitsuba 2: https://github.com/mitsuba-renderer/mitsuba2
  4. Michael Reiter, Marco Erler, Christoph Kuhn, Christian Gusenbauer, Johann Kastner. 2016. SimCT: a simulations tool for X-ray imaging, Proceedings of Conference on Industrial Computed Tomography (iCT2016), Wels, Austria, pp. 7.
  5. Artem Amirkhanov, Christoph Heinzl, Michael Reiter, Johann Kastner, Meister Eduard Gröller. 2011. Projection-based metal-artifact reduction for industrial 3D X-ray computed tomography. Visualization and Computer Graphics, IEEE Transactions on 17 (12), 2193-2202.

Responsible

For more information please contact Eduard Gröller, Christoph Heinzl , Thomas Auzinger .