Faculty of Science, Mohammed V University in Rabat, Rabat, Morocco.
Phys Med Biol. 2018 Apr 13;63(8):085008. doi: 10.1088/1361-6560/aab7a1.
The use of phase space in medical linear accelerator Monte Carlo (MC) simulations significantly improves the execution time and leads to results comparable to those obtained from full calculations. The classical representation of phase space stores directly the information of millions of particles, producing bulky files. This paper presents a virtual source model (VSM) based on a reconstruction algorithm, taking as input a compressed file of roughly 800 kb derived from phase space data freely available in the International Atomic Energy Agency (IAEA) database. This VSM includes two main components; primary and scattered particle sources, with a specific reconstruction method developed for each. Energy spectra and other relevant variables were extracted from IAEA phase space and stored in the input description data file for both sources. The VSM was validated for three photon beams: Elekta Precise 6 MV/10 MV and a Varian TrueBeam 6 MV. Extensive calculations in water and comparisons between dose distributions of the VSM and IAEA phase space were performed to estimate the VSM precision. The Geant4 MC toolkit in multi-threaded mode (Geant4-[mt]) was used for fast dose calculations and optimized memory use. Four field configurations were chosen for dose calculation validation to test field size and symmetry effects, [Formula: see text] [Formula: see text], [Formula: see text] [Formula: see text], and [Formula: see text] [Formula: see text] for squared fields, and [Formula: see text] [Formula: see text] for an asymmetric rectangular field. Good agreement in terms of [Formula: see text] formalism, for 3%/3 mm and 2%/3 mm criteria, for each evaluated radiation field and photon beam was obtained within a computation time of 60 h on a single WorkStation for a 3 mm voxel matrix. Analyzing the VSM's precision in high dose gradient regions, using the distance to agreement concept (DTA), showed also satisfactory results. In all investigated cases, the mean DTA was less than 1 mm in build-up and penumbra regions. In regards to calculation efficiency, the event processing speed is six times faster using Geant4-[mt] compared to sequential Geant4, when running the same simulation code for both. The developed VSM for 6 MV/10 MV beams widely used, is a general concept easy to adapt in order to reconstruct comparable beam qualities for various linac configurations, facilitating its integration for MC treatment planning purposes.
在医学线性加速器蒙特卡罗(MC)模拟中使用相空间可显著提高执行时间,并得到与全计算结果相当的结果。经典相空间表示直接存储数百万个粒子的信息,生成庞大的文件。本文提出了一种基于重建算法的虚拟源模型(VSM),该模型以从国际原子能机构(IAEA)数据库中提供的相空间数据自由生成的大约 800kb 的压缩文件作为输入。该 VSM 包括两个主要组件;初级和散射粒子源,并为每个源开发了特定的重建方法。从 IAEA 相空间中提取能谱和其他相关变量,并将其存储在两个源的输入描述数据文件中。对三种光子束(Elekta Precise 6 MV/10 MV 和 Varian TrueBeam 6 MV)对 VSM 进行了验证。在水中进行了广泛的计算,并对 VSM 和 IAEA 相空间的剂量分布进行了比较,以估计 VSM 的精度。在多线程模式(Geant4-[mt])下使用了 Geant4 MC 工具包进行快速剂量计算和优化内存使用。选择了四个场配置进行剂量计算验证,以测试场大小和对称性效应,正方形场的 [Formula: see text] [Formula: see text],[Formula: see text] [Formula: see text] 和 [Formula: see text] [Formula: see text],非对称矩形场的 [Formula: see text] [Formula: see text]。对于每个评估的辐射场和光子束,在单个工作站上进行 60 小时的计算时间内,以 3%/3 mm 和 2%/3 mm 标准,获得了基于[Formula: see text]形式的良好一致性,对于所有研究的情况,使用协议区概念(DTA)分析 VSM 在高剂量梯度区域的精度,也得到了令人满意的结果。在所有研究的情况下,在构建区和半影区中,平均 DTA 小于 1 mm。在计算效率方面,当对两者使用相同的模拟代码时,与顺序 Geant4 相比,使用 Geant4-[mt]的事件处理速度快六倍。广泛用于 6 MV/10 MV 束的开发的 VSM 是一种通用概念,易于适应各种直线加速器配置,以重建可比的束质量,方便其集成用于 MC 治疗计划目的。