Fraunhofer Institute for Integrated Circuits IIS
Fraunhofer Institute for Integrated Circuits IIS

BeyondVoxels – A Paradigm Shift in 3D Data Analysis for X-ray Computed Tomography

BeyondVoxels is developing a new, fast method for processing massive 3D X-ray data in the field of X-ray computed tomography. It is a joint project of the University of Passau, the Helmholtz-Zentrum Hereon, and Fraunhofer IIS. The core approach is a hierarchical wavelet-based access method that significantly reduces computation time and memory requirements and runs on standard off-the-shelf hardware. The solution integrates all steps of data analysis – from preprocessing and machine-assisted image recognition to interactive visualization – into a continuous pipeline, thereby enabling the visualization and analysis of CT datasets in the terabyte range.

Details zu BeyondVoxels

Expand all Close all

© Fraunhofer IIS
View through polarized glasses of a stereoscopically rendered 3D computed tomography dataset.
  • Cultural heritage: Exploration, documentation, and analysis of cultural objects.
  • Geology and materials: Investigation of geological, material, and sample structures.
  • Industry: Component analysis and quality control, inspection of workpieces.
  • Advantage: Larger CT datasets can be analyzed more efficiently without the need to purchase expensive specialized hardware.

  • Analogous to a pixel in 2D images, a voxel describes the three-dimensional image element of a CT dataset.
  • BeyondVoxels replaces conventional voxel-based approaches with a multi-level, hierarchical wavelet analysis that selectively loads detailed information as needed.
  • The pipeline supports scalable visualizations that enable users to gain rapid insights into complex 3D structures.

  • Partners: University of Passau, Fraunhofer IIS, and Helmholtz-Zentrum Hereon.
  • Project leadership: University of Passau – Prof. Dr. Tomas Sauer (Head of the FORWISS Institute) and Prof. Dr. Christoph Heinzl (Chair of Cognitive Sensor Systems; Fraunhofer IIS). External participation by Helmholtz-Zentrum Hereon – Dr. Julian Moosmann.
  • Funding: ErUM-Data, a funding program of the Federal Ministry of Research, Technology and Space. Duration: 3 years. Thematic focus: software and algorithms with an emphasis on artificial intelligence and machine learning, research data management, and federated digital infrastructures.