Hybrid OpenMP/MPI Monte Carlo Simulations in PET

• Main Authors: Hsiang Yun Wei, 魏湘芸
• Other Authors: Wei Min Jeng
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/87362909281733536796

碩士 === 東吳大學 === 資訊科學系 === 92 === Positron Emission Tomography (PET) is a technology that has useful applications in both the clinical and research settings. It uses a compound that is labeled with radioactivity that allows the imaging and measuring of biological processes. Monte Carlo simulation of three-dimensional PET data is critical in understanding the performances of image reconstruction algorithms for corresponding instrumentations. The random numbers used in sampling values from appropriate frequency functions is central and the generator implementation plays an important part. The aim of this research is to reduce the computing time by parallelizing the simulation on a cluster of symmetric multiple processors. Both shared-memory and message passing programming paradigms are employed to obtain the high throughput to expedite the simulation process. Parallelization strategy utilized in the constructs demonstrated its superb performance over prior work.