Implementation of 3D Monte Carlo PET reconstruction algorithm on GPU
Department of Diagnostic Radiology & Oncotherapy, Semmelweis University, Budapest, Hungary
IEEE Nuclear Science Symposium Conference Record (NSS/MIC), 2009
@inproceedings{wirth2009implementation,
title={Implementation of 3D Monte Carlo PET reconstruction algorithm on GPU},
author={Wirth, A. and Cserkaszky, A. and K{‘a}ri, B. and L{‘e}gr{‘a}dy, D. and Feh{‘e}r, S. and Czifrus, S. and Domonkos, B.},
booktitle={Nuclear Science Symposium Conference Record (NSS/MIC), 2009 IEEE},
pages={4106–4109},
year={2009},
organization={IEEE}
}
The presentation summarizes our achieved results within the framework of the cross-institutional research project called TeraTomo. The aim of the latter is to create a high-speed algorithm optimized on graphics processors (GPU) providing more accurate PET reconstructions by taking advantage from CT data. The current paper discusses a hit-and-miss Monte Carlo approach as an “on-the-fly” forward projector of an ML-EM algorithm.
May 19, 2011 by hgpu
