Department of Radiation Oncology, William Beaumont Hospital, Royal Oak, MI, 48073, USA.
Department of Medical Physics, Wayne State University, Detroit, MI, 48202, USA.
Med Phys. 2018 Jul;45(7):2864-2874. doi: 10.1002/mp.12926. Epub 2018 May 11.
A treatment planning/delivery QA tool using linac log files (LF) and Monte Carlo (MC) dose calculation is investigated as a standalone alternative to phantom-based patient-specific QA (ArcCHECK (AC)).
Delivering a variety of fields onto MapCHECK2 and ArcCHECK, diode sensitivity dependence on dose rate (in-field) and energy (primarily out-of-field) was quantified. AC and LF QAs were analyzed with respect to delivery complexity by delivering 12 × 12 cm static fields/arcs comprised of varying numbers of abutting sub-fields onto ArcCHECK. About 11 clinical dual-arc VMAT patients planned using Pinnacle's convolution-superposition (CS) were delivered on ArcCHECK and log file dose (LF-CS and LF-MC) calculated. To minimize calculation time, reduced LF-CS sampling (1/2/3/4° control point spacing) was investigated. Planned ("Plan") and LF-reconstructed CS and MC doses were compared with each other and AC measurement via statistical [mean ± StdDev(σ)] and gamma analyses to isolate dosimetric uncertainties and quantify the relative accuracies of AC QA and MC-based LF QA.
Calculation and ArcCHECK measurement differed by up to 1.5% in-field due to variation in dose rate and up to 5% out-of-field. For the experimental segment-varying plans, despite CS calculation deviating by as much as 13% from measurement, Plan-MC and LF-MC doses generally matched AC measurement within 3%. Utilizing 1° control point spacing, 2%/2 mm LF-CS vs AC pass rates (97%) were slightly lower than Plan-CS vs AC pass rates (97.5%). Utilizing all log file samples, 2%/2 mm LF-MC vs AC pass rates (97.3%) were higher than Plan-MC vs AC (96.5%). Phantom-dependent, calculation algorithm-dependent (MC vs CS), and delivery error-dependent dose uncertainties were 0.8 ± 1.2%, 0.2 ± 1.1%, and 0.1 ± 0.9% respectively.
Reconstructing every log file sample with no increase in computational cost, MC-based LF QA is faster and more accurate than CS-based LF QA. Offering similar dosimetric accuracy compared to AC measurement, MC-based log files can be used for treatment planning QA.
研究一种使用直线加速器日志文件（LF）和蒙特卡罗（MC）剂量计算的治疗计划/交付 QA 工具，作为基于体模的患者特定 QA（ArcCHECK（AC））的独立替代方案。
在 MapCHECK2 和 ArcCHECK 上测量各种射野，量化二极管剂量率（场内）和能量（主要是场外）依赖性。通过在 ArcCHECK 上交付由不同数量相邻子野组成的 12×12cm 静态射野/弧形，分析 AC 和 LF QA 相对于交付复杂性。使用 Pinnacle 的卷积叠加（CS）为大约 11 个临床双弧 VMAT 患者进行计划，并在 ArcCHECK 上进行交付，并计算日志文件剂量（LF-CS 和 LF-MC）。为了最小化计算时间，研究了 LF-CS 的减少采样（1/2/3/4°控制点间距）。通过统计[平均值±标准差（σ）]和伽马分析将计划（“Plan”）和 LF 重建的 CS 和 MC 剂量与 AC 测量进行比较，以隔离剂量不确定性并量化 AC QA 和基于 MC 的 LF QA 的相对准确性。
由于剂量率的变化，场内计算和 ArcCHECK 测量的差异最大可达 1.5%，场外最大可达 5%。对于实验分段变化的计划，尽管 CS 计算与测量的差异最大可达 13%，但 Plan-MC 和 LF-MC 剂量通常与 AC 测量相差在 3%以内。使用 1°控制点间距，LF-CS 与 AC 的 2%/2mm 通过率（97%）略低于 Plan-CS 与 AC 的通过率（97.5%）。使用所有日志文件样本，LF-MC 与 AC 的 2%/2mm 通过率（97.3%）高于 Plan-MC 与 AC 的通过率（96.5%）。与体模相关、与计算算法相关（MC 与 CS）以及与交付误差相关的剂量不确定性分别为 0.8±1.2%、0.2±1.1%和 0.1±0.9%。
在不增加计算成本的情况下重建每个日志文件样本，基于 MC 的 LF QA 比基于 CS 的 LF QA 更快、更准确。与 AC 测量相比，提供类似的剂量准确性，基于 MC 的日志文件可用于治疗计划 QA。
