MENU

Click here to

Ă—

Are you sure ?

Yes, do it No, cancel

Implementing a Phase Space File Framework in MCGPU

R Trevisan Massera1,2*, R Thomson2, A Tomal1, (1) University of Campinas, BR, (2) Carleton Univ, Ottawa, ON

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: To implement and validate a phase space files algorithm in MCGPU, a Monte Carlo (MC) code developed for graphical processing units (GPU). We propose to take advantage of the fast simulation speeds for photon simulations in GPUs by splitting complex multiscale MC simulations into different steps, improving the efficiency of these simulations.


Methods: A ray-tracing algorithm adapted from PENELOPE was implemented to work with MCGPU voxelized geometries to generate PSF data for up to 5 ROIs. Afterwards, the output binary files are converted to the IAEA standard. We compared the phase space file particle energy spectrum and voxel glandular dose for different breast imaging techniques: mammography (W/Rh 28 kV), tomosynthesis (W/Al 31 kV), contrast enhanced digital mammography (W/Cu 49 kV) and breast-CT (W/Al 49 kV) acquisitions. An anthropomorphic computer-generated breast model was used. The energy spectrum and voxel glandular dose were compared with the ones obtained from PENELOPE 2018 + penEasy 2019 and egs_brachy (EGSnrc) MC simulations.


Results: The particle energy spectra extracted from the phase space file show excellent agreement with other MC codes, with average energies differing by less than 0.2%. The voxel doses from MCGPU agree with PENELOPE and egs_brachy doses loaded from phase space files within 0.6%, which are consistent with 2 standard deviations of the statistical uncertainty. Moreover, in our tests, MCGPU presented speed gains up to 100 times compared to PENELOPE and EGSnrc.


Conclusion: The framework introduces the option to generate phase space files in MCGPU and allows its usage for any MC code that follows the IAEA PSF standard. This can be advantageous to improve the efficiency of the simulations, where macroscopic scale would be handled by MCGPU and the small detailed scale in another code that includes electron transport in microscopic energy deposition studies. Our implementations will be freely distributed online.

Funding Support, Disclosures, and Conflict of Interest: Funding support from: Natural Sciences and Engineering Research Council (NSERC) of Canada; Ontario Ministry of Research and Innovation; Canada Research Chairs; Conselho Nacional de Desenvolvimento Cientifico e Tecnologico (CNPQ) project 140155/2019-8; FAPESP project 2015/21873-8; ELAP scholarship with support of the Government of Canada; AAPM International Training and Research Coordination scholarship.

Keywords

Not Applicable / None Entered.

Taxonomy

IM- Breast X-Ray Imaging: Monte Carlo modeling

Contact Email