Center for Visual Computing, Computer Science Department, Stony Brook University, Stony Brook, NY 11794

Abstract

The task of reconstructing an object from its projections via tomographic methods is a time-consuming process due to the vast complexity of the data. For this reason, manufacturers of equipment for medical computed tomography (CT) rely mostly on special ASICs to obtain the fast reconstruction times required in clinical settings. Although modern CPUs have gained sufficient power in recent years to be competitive for 2D reconstruction, this is not the case for 3D reconstructions, especially not when iterative algorithms must be applied. Incidentally, this has prevented some very effective algorithms to be applied in clinical practice, as well as in general research. However, the recent evolution of commodity PC computer graphics boards (GPUs) has the potential to change this picture in a very dramatic way. We have shown that the new GPUs can be exploited to perform both analytical and iterative reconstruction from X-ray and functional imaging data at clinical rates and high quality. We have decomposed three popular 3D reconstruction algorithms into a common set of base modules, which all can be executed on the GPU and their output linked internally. The data never leave the GPU, which eliminates previous costly GPU-CPU bottlenecks. Visualization of the reconstructed object is easily achieved since the object already resides in the graphics hardware, allowing one to run a visualization module at any time to view the reconstruction results. Our implementation allows speedups of 1-2 orders of magnitude over software implementations, at comparable image quality.

Downloads

[papers]

• Fang Xu and Klaus Mueller, "Towards a unified framework for rapid 3D computed tomography on commodity GPUs," IEEE Medical Imaging Conference'03, Portland, OR, 2003.
• Fang Xu and Klaus Mueller, "Ultra-fast 3D filtered backprojection on commodity graphics hardware," IEEE International Symposium on Biomedical Imaging'04, Washington D.C., 2004.
• Fang Xu and Klaus Mueller, "RapidCT: Acceleration of 3D computed tomography on the GPU," ACM Workshop on General-Purpose Computing on Graphics Processors 2004 (GPGPU'04), Los Angeles, August 2004.
• Fang Xu and Klaus Mueller, "Accelerating popular tomographic reconstruction algorithms on commodity PC graphics hardware," to appear on IEEE Transaction of Nuclear Science, 2005.

[slides]

• Poster on IEEE Medical Imaging Conference, 2003 (pdf)
• Poster on IEEE International Symposium on Biomedical Imaging, 2004 (pdf)
  

Configurations

• Platform: Pentium 4 2.66GHz PC, 512MB RAM, GeForce FX 5900 (256MB RAM)
• Experiment Data: 3D Shepp Logan Phantom
• contrast levels: 0.5%, 1% and 2%
• resolutions: 128³ and 256³
• cone angle: 15 degrees
• Algorithms:
• SART (Simultaneous Algebraic Reconstruction Technique)
• ML-EM/OS-EM (Maximum Likelihood - Expectation Maximization / Ordered-Subset Expectation Maximization)
• FDK-FBP (Feldkamp-Davis-Kress Filtered Backprojection)
• Supported Geometries:
• Cone-beam
• Fan-beam
• Parallel-beam