Premium
SU‐FF‐I‐50: Whole Body and Distal Organ‐Specific Dosimetry Using Parallel SN Methods
Author(s) -
AlBasheer A,
Ghita M,
Sjoden G,
Bolch W,
Lee C
Publication year - 2007
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.2760427
Subject(s) - imaging phantom , monte carlo method , dosimetry , voxel , physics , computation , code (set theory) , computational science , medical imaging , radiation transport , computer science , algorithm , nuclear medicine , mathematics , optics , statistics , artificial intelligence , medicine , set (abstract data type) , programming language
Purpose: To create a new methodology for dose computations applicable to general medical physics applications, based upon a direct deterministic solution of the (3‐D) Boltzmann transport equation (BTE). Method and Materials: Using the multigroup discrete ordinates (S N ) deterministic method in the PENTRAN‐MP (Parallel Environment Neutral‐particle TRANsport‐Medical Physics) code system, fluxes and corresponding doses were determined for a clinical test case using a UF Series B whole‐body voxelized pediatric patient phantom. With a 90 keV planar x‐ray source, we performed two sets of transport calculations based on the same phantom model, employing both the Monte Carlo (MC) and S N methods. S N calculations were performed using PENTRAN, and the MCNP5 code was used for MC calculations. Post processing in the PENTRAN‐MP code system includes seamless parallel data extraction using the PENDATA code, followed by application of the 3D‐DOSE code to compute dose in each phantom voxel, with dose‐volume histograms for critical organs of interest. Results: We demonstrate that the deterministic S N and MC solutions are, in the vicinity of the source region, completely in agreement; the largest statistical error of the MC simulation was <5%, typically averaging <2%. Moreover, the deterministic S N method also provided globally converged results in each voxel at sites distal to the source (in ∼3 hrs on 16 processors), where the MC results on the same computing platform had significant out of field error, and would require a significantly longer running time (estimated at >2000 hrs) to produce meaningful results with an acceptable stochastic error to enable comparison with the well‐converged S N results. Conclusions: A new methodology for 3D dose calculations has been developed based on a parallel discrete ordinates (S N ) solution of the BTE. With the proper discretization applied, the S N method presents an accurate, fast solution yielding a complete 3‐D dose distribution without stochastic error.