虚拟源模型方法是使用解析方法建立虚拟源模型，并结合蒙特卡罗方法进行剂量计算的源模型方法，可以同时发挥解析方法与蒙卡方法的优势，在临床治疗计划的剂量验证以及加速器的commission过程中都可以发挥作用。基于蒙卡的剂量验证在临床上不仅可以精确完成剂量分布计算，同时可以避免基于测量的剂量验证方式额外需要的机器耗损。本文使用虚拟源模型方法，将加速器出束粒子分为三类：初始光子、散射光子和污染电子。根据水箱中的百分深度剂量与空气中的输出因子，对初始光子源和散射光子源进行建模，得到其能量和方向分布。分别对初始光子源和散射光子源进行采样，得到单个光子的能量与方向，同时对光子的能量和权重进行相应的修正。将所有的光子导入快速蒙卡计算软件中，计算水模体中的剂量沉积，验证虚拟源模型方法的准确性。在4个不同大小射野的模拟数据中，PDD误差基本控制在±1.5%以内，未考虑电子的影响，浅部数据的误差影响较大，当深度小于3cm时，误差会超过5%；对每一个特定大小的射野，5个不同深度处的OAR，射野内的误差均在±2%以内，射野边缘由于半影区的存在，剂量数值误差较大，使用位置误差DTA分析，基本在±2mm以内。将虚拟源模型应用于三台不同型号的直线加速器，模拟与测量数据之间的比较与第一台直线加速器的结果相类似，误差均在可接受范围内。相对于全部使用水箱测量数据进行加速器的commission，使用虚拟源模型方法可以更快地完成这一工作，为加速器建模和自动的commission提供了新的思路。解析计划系统中导出的Dicom文件，获得实际病例治疗的加速器运行数据，并将病人CT数据转化为密度模体，使用虚拟源模型方法与GPU并行计算模拟病人体内的剂量分布，与TPS进行比较，作为第三方的剂量验证。剂量分析采用DVH和Gamma分析两种评估方法，对应于Pinnacle系统和Eclipse系统计划的通过率均大于95%。
The virtual source model is to use the analytical method to establish the virtual source model, and combined with the Monte Carlo method for the dosecalculation process, which can simultaneously play the advantages of the analytical method and the Monte Carlo method, which can be used in the doseverification of the clinical treatment plan and the commissioning process of theaccelerator Play a role.This paper use the virtual source model method to classify the accelerator beam particles into three categories: initial photon source, scattered photon source and polluted electron source. In the process of particle sampling, the initial photon source and the scattered photon source are sampled to obtain the energy and direction information of the photon, and the energy and the weight of the photon are corrected accordingly. In the four different sizes of field simulation data, in the 5cm to 25cm PDD errors are basically controlled within ± 1.5%, without considering the impact of electrons, shallow data error is greater, when the depth is less than 3cm, the error Will be more than 5%. For each specific size of the field, respectively, compared the depth of 5 OAR, within the shooting field within the error are within ± 2%, shooting field edge due to the presence of penumbra, the dose value error is large. The virtual source model is applied to another linear accelerator. The simulated data and the measured data error of the accelerator are controlled within the allowable range.The Dicom file derived from the planning system was used to obtain the accelerator operation data of the actual treatment, and the patient CT data was transformed into the density model. The dose distribution was compared and tested as a third party dose. Dose comparison using DVH and Gamma analysis of two methods, corresponding to the Pinnacle system and Eclipse system plan pass rate is greater than 95%. The virtual source model can be used to model the accelerator beam out and perform third-party dose verification.
