常规QA设备对脊柱转移瘤SRS/SBRT计划的验证及存在问题探讨

Dosimetric verification of SRS/SBRT planning for spinal metastases using conventional QA devices and discussion of existing problems

目的:对比aSi-1200 EPID与Octavius 1500电离室矩阵在6 MV非均整脊柱转移瘤计划验证中的应用,探讨可能存在的问题。方法:23例计划在全局归一方式下以不同标准和阈值行2Dγ通过率(GPR)评估,将治疗计划系统(TPS)中患者数据导入Verisoft软件重建剂量。引入不同类型和量值的误差:MLC的透射率(TF)和剂量叶片间隙(DLG)、治疗等中心和MU误差,使用GPR方法量化Octavius 1500识别这些误差的灵敏度,并通过评估靶区和危及器官(OARs)剂量学指标变化研究TF和DLG误差的临床意义。结果:与Octavius 1500矩阵相比,EPID方式在同一标准不同阈值下GPR均值更大数据离散程度更小,重建与计算的剂量体积直方图分析表明临床靶区和计划靶区D_(min)、D_(max)和D_(mean)的百分剂量偏差(DD%)较大,其中临床靶区最大DD%值为50.00%、11.31%和-8.71%,计划靶区最大DD%值为-25.86%、9.31%和-8.22%。治疗等中心和+3%、+5%的MU误差都被检测到,所有标准均未检测到TF误差,+0.3 mm的DLG误差只被2%/3 mm的标准检测到。TF值增大0.0236%模型误差和DLG增加0.3 mm模型误差均导致靶区和OARs的剂量增加,其中OARs剂量增加较明显,尤其健侧肺的V_(20)分别增加9.80%和8.85%,脊髓D_(0.1 cc)均增加5.35%。结论:使用GPR方法识别引入误差的MLC模型可靠性不够,提示鉴别TPS问题的根源需要更有效的独立质量保证方法以确保通过TPS计算得出的陡峭剂量梯度的可信度。剂量验证系统软件算法的独立验证研究是有必要的,以确定算法局限性导致的重建剂量不确定性范围。

Objective To compare the applications of asi-1200 EPID and Octavius 1500 matrix in the dosimetric verification for 6 MV flattening filter free plan of spinal metastases,and to explore potential problems.Methods Under global normalization,a total of 23 plans were evaluated by 2D gamma passing rates(GPR)at different criteria with different thresholds,and the patient data from treatment planning system(TPS)were imported into Verisoft software for dose reconstruction.Errors of different types and magnitudes including the errors from MLC transmission factor(TF),dosimetric leaf gap(DLG),treatment iso-center and MU were introduced.The sensitivity of Octavius 1500 matrix indentifying these errors was quantified by GPR method,and the clinical significance of TF and DLG errors was studied by evaluating the dosimetric deviations of target areas and organs-at-risk(OAR).Results Compared with Octavius 1500 matrix,EPID had a higher mean GPR at the same criteria with different thresholds and achieved a smaller dispersion degree of data.The dosevolume histogram(DVH)analysis between reconstruction and calculation showed that the percentage dose differences(DD%)of the minimum dose(D_(min)),maximum dose(D_(max))and mean dose(D_(mean))to clinical target volume(CTV)and planning target volume(PTV)were relatively larger,in which the maximum DD%of CTV were 50.00%,11.31%and 8.71%,respectively,and the maximum DD%of PTV were-25.86%,9.31%and-8.22%,respectively.Errors from treatment isocentor and MU of+3%,+5%could be detected.TF error couldn't be detected at all criteria,and only+0.3 mm DLG error was detectable at 2%/3 mm criteria.A 0.0236%increase of TF and+0.3 mm DLG error resulted in dose increase to target areas and OAR,among which the dose increase of OAR was obvious,especially V_(20) of the healthy lungs which was increased by 9.80%and 8.85%,and D_(0.1 cc) to the spinal cord which was increased by 5.35%.Conclusion The inadequate reliability of using GPR to distinguish error-introducing MLC models suggests that more effective independent quality assurance methods are needed to identify the root cause of TPS problems,thereby ensuring the confidence of steep dose gradients calculated by TPS.The independent verification research on software algorithms of the dosimetric verification system is necessary for determining the uncertainty of reconstructed dose due to algorithm limits.