The research on x-ray computed tomography is an interconnected challenge of system design, hardware optimization and image reconstruction techniques. Advances and new developments in one field influence developments in another. Therefore we are well positioned in 3D-reconstruction techniques, artefact and misalignment correction, as well as in geometric calibration. The focus of our research is in improvements for intra-operative scenarios.
You are here:
One of our main competences is research on 3D-reconstruction techniques. We developed a method to implement arithmetic and algebraic reconstruction techniques for use in high-performance heterogeneous environments. Both implementations take advantage of the parallel processing power of modern GPUs to provide exceptional quality in very short time.
The implemented filtered back-projection algorithm allows rapid reconstruction of very large volumes. A special mode allows simultaneous image acquisition and reconstruction.
Our implementation of the simultaneous arithmetic reconstruction technique (SART) allows a flexible balance between reconstruction quality, resolution and runtime. It supports heterogeneous multi-GPU architectures for the optimal resource utilization and was designed for the use with freely definable trajectories opening new imaging geometries.
• Filtered Back-Projection FBP
• Simultaneous Arithmetic Reconstruction Technique SART
• Multi-GPU Architecture
These topics have been developed within the BMBF supported Project ORBIT - grant 13EZ1115B, initially conducted from 2010 to 2014 at Fraunhofer and Charité and since by Charité.
Contact: Erwin Keeve