Department of Radiation Oncology, Medical University of Vienna, Vienna, Austria.
Faculty of Engineering, University of Applied Sciences, Wiener Neustadt, Austria.
Med Phys. 2021 Aug;48(8):4560-4571. doi: 10.1002/mp.14987. Epub 2021 Jun 28.
In the past years, many different neural network-based conversion techniques for synthesizing computed tomographys (sCTs) from MR images have been published. While the model's performance can be checked during the training against the test set, test datasets can never represent the whole population. Conversion errors can still occur for special cases, for example, for unusual anatomical situations. Therefore, the performance of sCT conversion needs to be verified on a patient specific level, especially in the absence of a planning CT (pCT). In this study, the capability of cone-beam CTs (CBCTs) for the validation of sCTs generated by a neural network was investigated.
41 patients with tumors in the head region were selected. 20 of them were used for model training and 10 for validation. Different implementations of CycleGAN (with/without identity and feature loss) were used to generate sCTs. The pixel (MAE, RMSE, PSNR) and geometric error (DICE, Sensitivity, Specificity) values were reported to identify the best model. VMAT plans were created for the remaining 11 patients on the pCTs. These plans were re-calculated on sCTs and CBCTs. An automatic density overriding method ( ) and a population-based dose calculation method ( ) were employed for CBCT-based dose calculation. The dose distributions were analysed using 3D global gamma analysis, applying a threshold of 10% with respect to the prescribed dose. Differences in DVH metrics for the PTV and the organs-at-risk were compared among the dose distributions based on pCTs, sCTs, and CBCTs.
The best model was the CycleGAN without identity and feature matching loss. Including the identity loss led to a metric decrease of 10% for DICE and a metric increase of 20-60 HU for MAE. Using the 2%/2 mm gamma criterion and pCT as reference, the mean gamma pass rates were 99.0 0.4% for sCTs. Mean gamma pass rate values comparing pCT and CBCT were 99.0 0.8% and 99.1 0.8% for the and , respectively. The mean gamma pass rates comparing sCT and CBCT resulted in 98.4 1.6% and 99.2 0.6% for and , respectively. The differences between the gamma-pass-rates of the sCT and two CBCT-based methods were not significant. The majority of deviations of the investigated DVH metrices between sCTs and CBCTs were within 2%.
The dosimetric results demonstrate good agreement between sCT, CBCT, and pCT based calculations. A properly applied CBCT conversion method can serve as a tool for quality assurance procedures in an MR only radiotherapy workflow for head patients. Dosimetric deviations of DVH metrics between sCT and CBCTs of larger than 2% should be followed up. A systematic shift of approximately 1% should be taken into account when using the approach in an MR only workflow.
在过去的几年中，已经发表了许多基于神经网络的转换技术，用于从磁共振图像（MR 图像）合成计算机断层扫描（sCT）。虽然可以在训练过程中针对测试集检查模型的性能，但测试数据集永远无法代表整个人群。对于特殊情况（例如不常见的解剖情况），转换错误仍可能发生。因此，需要在特定患者的基础上验证 sCT 转换的性能，特别是在没有计划 CT（pCT）的情况下。在这项研究中，研究了锥形束 CT（CBCT）在验证神经网络生成的 sCT 方面的能力。
选择了 41 名头部肿瘤患者。其中 20 名用于模型训练，10 名用于验证。使用不同的 CycleGAN 实现（有无身份和特征损失）生成 sCT。报告像素（MAE、RMSE、PSNR）和几何误差（DICE、灵敏度、特异性）值以确定最佳模型。在 pCT 上为其余 11 名患者创建 VMAT 计划。在 sCT 和 CBCT 上重新计算这些计划。使用自动密度覆盖方法（ ）和基于人群的剂量计算方法（ ）进行 CBCT 剂量计算。使用 3D 全局伽马分析对剂量分布进行分析，应用相对于规定剂量的 10%的阈值。比较基于 pCT、sCT 和 CBCT 的剂量分布中 PTV 和危及器官的 DVH 指标之间的差异。
最佳模型是没有身份和特征匹配损失的 CycleGAN。包含身份损失会导致 DICE 下降 10%，MAE 增加 20-60 HU。使用 2%/2 mm 伽马标准和 pCT 作为参考，sCT 的平均伽马通过率为 99.0 ± 0.4%。比较 pCT 和 CBCT 的平均伽马通过率值分别为 99.0 ± 0.8%和 99.1 ± 0.8% 和 ，分别。比较 sCT 和 CBCT 的平均伽马通过率值分别为 98.4 ± 1.6%和 99.2 ± 0.6% 和 ，分别。sCT 和两种基于 CBCT 的方法的伽马通过率之间的差异没有统计学意义。DVH 指标的大多数偏差在 sCT 和 CBCT 之间都在 2%以内。
剂量学结果表明，sCT、CBCT 和 pCT 之间的计算结果具有良好的一致性。在仅使用磁共振成像（MR）的放射治疗流程中，正确应用 CBCT 转换方法可以作为质量保证程序的工具。sCT 和 CBCT 之间的 DVH 指标的剂量学偏差应超过 2%。在仅使用 MR 的工作流程中使用 方法时，应考虑约 1%的系统偏移。
