Division of Medical Radiation Physics, Department of Radiation Oncology, Inselspital, Bern University Hospital and University of Bern, Bern, Switzerland.
Elekta Brachytherapy, 3905TH, Veenendaal, The Netherlands.
Med Phys. 2017 Oct;44(10):5475-5485. doi: 10.1002/mp.12490. Epub 2017 Aug 31.
Most dose calculations for HDR brachytherapy treatments are based on the AAPM-TG43 formalism. Because patient's anatomy, heterogeneities, and applicator shielding are not considered, the dose calculation based on this formalism is inaccurate in some cases. Alternatively, collapsed cone (CC) methods as well as Monte Carlo (MC) algorithms belong to the model-based dose calculation algorithms, which are expected to improve the accuracy of calculated dose distributions. In this work, the performance of a CC algorithm, ACE in Oncentra Brachy 4.5 (ACE 4.5), has been investigated by comparing the calculated dose distributions to the AAPM-TG43 and MC calculations for 10 HDR brachytherapy accelerated partial breast irradiation treatments (APBI). Comparisons were also performed with a corrected version of ACE 4.5 (ACE 4.5/corr).
The brachytherapy source microSelectron mHDR-v2 (Elekta Brachytherapy) has been implemented in a MC environment and validated by comparing MC dose distributions simulated in a water phantom of 80 cm in diameter with dose distributions calculated with the AAPM-TG43 algorithm. Dose distributions calculated with ACE 4.5, ACE 4.5/corr, AAPM-TG43 formalism, and MC for 10 APBI patients plans have then been computed and compared using HU scaled densities. In addition, individual dose components have been computed using ACE 4.5, ACE 4.5/corr, and MC, and compared individually.
Local differences between MC and AAPM-TG43 calculated dose distributions in a large water phantom are < 1%. When using HUs scaled densities for the breast cancer patients, both accuracy levels of ACE 4.5 overestimate the MC calculated dose distributions for all analyzed dosimetric parameters. In the planning target volume (PTV), ACE 4.5 (ACE 4.5/corr) overestimates on average V by 3% ± 1% (1% ± 1%) and D by 3% ± 1% (1% ± 1%) and in the organs at risk D by 4% ± 2% (1% ± 1%), D by 4% ± 2% (0% ± 1%), and D by 8% ± 2% (3% ± 1%) compared to MC. Comparisons of the individual dose components reveals an agreement for the primary component of < 2% local differences for both ACE 4.5 and ACE 4.5/corr. Local differences of about 40% (20%) for the first and residual scatter components where observed when using ACE 4.5 (ACE 4.5/corr). Using uniform densities for one case shows a better agreement between ACE 4.5 and MC for all dosimetric parameters considered in this work.
In general, on the 10 APBI patients the ACE 4.5/corr algorithm results in similar dose distributions as the commonly used AAPM-TG43 within the PTV. However, the accuracy of the ACE 4.5/corr calculated dose distribution is closer to MC than to AAPM-TG43. The differences between commercial version ACE 4.5 and MC dose distributions are mainly located in the first and residual scatter components. In ACE 4.5/corr, the changes done in the algorithm for the scatter components substantially reduce these differences.
大多数 HDR 近距离治疗剂量计算都是基于 AAPM-TG43 形式主义。由于未考虑患者的解剖结构、异质性和施源器屏蔽，因此基于该形式主义的剂量计算在某些情况下并不准确。此外，锥形束（CC）方法和蒙特卡罗（MC）算法属于基于模型的剂量计算算法，预计会提高计算剂量分布的准确性。在这项工作中，通过将 ACE 4.5 计算的剂量分布与 AAPM-TG43 和 MC 计算的 10 个 HDR 近距离治疗加速部分乳房照射治疗（APBI）进行比较，研究了 CC 算法 ACE 的性能。还对 ACE 4.5 的修正版本（ACE 4.5/corr）进行了比较。
将微Selectron mHDR-v2 近距离治疗源（Elekta Brachytherapy）实现在 MC 环境中，并通过将直径为 80 厘米的水模中的 MC 模拟剂量分布与 AAPM-TG43 算法计算的剂量分布进行比较来验证。然后，使用 HU 缩放密度计算了 10 个 APBI 患者计划的 ACE 4.5、ACE 4.5/corr、AAPM-TG43 形式主义和 MC 计算的剂量分布，并进行了比较。此外，还使用 ACE 4.5、ACE 4.5/corr 和 MC 计算了各个剂量分量，并分别进行了比较。
在大水模中，MC 和 AAPM-TG43 计算的剂量分布之间的局部差异<1%。当使用乳腺癌患者的 HU 缩放密度时，ACE 4.5 的两个准确度级别都高估了 MC 计算的所有分析剂量参数的剂量分布。在计划靶区（PTV）中，ACE 4.5（ACE 4.5/corr）平均高估 V 为 3%±1%（1%±1%）和 D 为 3%±1%（1%±1%），在危及器官中 D 为 4%±2%（1%±1%），D 为 4%±2%（0%±1%），D 为 8%±2%（3%±1%）与 MC 相比。对个别剂量分量的比较表明，ACE 4.5 和 ACE 4.5/corr 的主要分量的局部差异小于 2%。当使用 ACE 4.5（ACE 4.5/corr）时，观察到第一和残余散射分量的局部差异约为 40%（20%）。对于一个病例使用均匀密度，在本工作中考虑的所有剂量参数方面，ACE 4.5 与 MC 之间的一致性更好。
一般来说，在 10 名 APBI 患者中，ACE 4.5/corr 算法在 PTV 内产生的剂量分布与常用的 AAPM-TG43 相似。然而，ACE 4.5/corr 计算的剂量分布的准确性比 AAPM-TG43 更接近 MC。商业版本 ACE 4.5 和 MC 剂量分布之间的差异主要位于第一和残余散射分量中。在 ACE 4.5/corr 中，对散射分量的算法更改大大减少了这些差异。
