4459

Acceleration of PET Monte Carlo simulation using the graphics hardware ray-tracing engine

Zhiguang Wang, Peter. D. Olcott, Craig S. Levin
Department of Radiology at Stanford University Medical School, 300 Pasteur Drive, Alway, M001, Stanford, CA 94305
IEEE Nuclear Science Symposium Conference Record (NSS/MIC), 2010

@inproceedings{wang2010acceleration,

   title={Acceleration of PET Monte Carlo simulation using the graphics hardware ray-tracing engine},

   author={Wang, Z. and Olcott, P. and Levin, C.S.},

   booktitle={Nuclear Science Symposium Conference Record (NSS/MIC), 2010 IEEE},

   pages={1848–1855},

   year={2010},

   organization={IEEE}

}

Source Source   

781

views

GRAY (High Energy Photon Ray Tracer) is a Monte-Carlo ray-driven high energy photon transport engine for PET and SPECT applications that supports complex mesh based primitives for source distributions, phantom shapes, and detector geometries. Ray tracing is a technique used in computer graphics to render scenes with realistic light properties. We adapted this technique to accelerate solving the intersection test problems in our simulator. Monte Carlo simulations study the performance of PET systems, test signal processing algorithms, guide the design of advanced PET systems, data correction schemes, and image reconstruction algorithms. GPU acceleration of these simulations makes these studies more practical while avoiding the need of a large, expensive computer cluster. Recent improvements in the computing power and programmability of graphics processing units (GPUs) have enabled the possibility of using GPUs for the acceleration of scientific applications, including time-consuming simulations in physics. This paper describes the acceleration of GRAY using NVIDIA OptiX ray tracing engine on GPUs for runtime performance and the implementation of secondary physical processes validated against GEANT for enhancing GRAY’s accuracy. We describe the GPU-based computation and how it is mapped onto the many parallel computational units now available on the NVIDIA GTX 200 series GPUs. For a brain PET acquisition benchmark, a speedup of 5.2X; for a GATE PET acquisition benchmark, a speedup of 15.88X were achieved on a single GTX285 GPU over the CPU GRAY and the GATE simulation toolkit executed on an Intel(R) Core2 Duo T6600@2.40GHz processor with equivalent accuracy.
No votes yet.
Please wait...

* * *

* * *

HGPU group © 2010-2017 hgpu.org

All rights belong to the respective authors

Contact us: