Learning-Based Stopping Power Mapping on Dual-Energy CT for Proton Radiation Therapy.
Learning-Based Stopping Power Mapping on Dual-Energy CT for Proton Radiation Therapy.
复制标题
DOI：
10.14338/ijpt-d-20-00020.1
复制
发表时间：
2021
影响因子：
1.7
通讯作者：
Yang X
中科院分区：
文献类型：
--
作者：
Wang T;Lei Y;Harms J;Ghavidel B;Lin L;Beitler JJ;McDonald M;Curran WJ;Liu T;Zhou J;Yang X
双能计算机断层扫描 (DECT) 已用于通过获取光子相互作用的能量依赖性来导出相对阻止本领 (RSP) 图。当使用基于物理的绘图技术时，DECT 导出的 RSP 地图可能会受到图像噪声水平和伪影严重程度的影响。这项工作提出了一种基于噪声稳健学习的方法，用于预测质子放射治疗 DECT 的 RSP 图。所提出的方法使用剩余注意周期一致的生成对抗网络，通过引入逆 RSP 到 DECT 映射，使 DECT 到 RSP 映射接近 1 对 1 映射。为了评估所提出的方法，我们回顾性研究了 20 名头颈癌患者在质子放射治疗模拟中进行 DECT 扫描。根据化学成分计算分配地面实况RSP值，并作为DECT数据集训练过程中的学习目标；使用留一交叉验证策略根据所提出的方法的结果对它们进行评估。预测的 RSP 图显示整个身体体积的平均归一化均方误差为 2.83%，所有感兴趣体积的平均误差小于 3%。在 DECT 数据集中添加额外的模拟噪声后，所提出的方法仍然保持了可比的性能，而基于物理的化学计量方法因噪声水平的增加而导致不准确度降低。对于 D95% 和 Dmax，临床靶体积的剂量体积直方图指标与地面真实值的平均差异小于 0.2 Gy，没有统计学意义。危险器官的剂量体积直方图指标的最大差异平均约为 1 Gy。这些结果强烈表明我们基于机器学习的方法预测的 RSP 图具有很高的准确性，并显示了其对于质子治疗计划和剂量计算的潜在可行性。
Dual-energy computed tomography (DECT) has been used to derive relative stopping power (RSP) maps by obtaining the energy dependence of photon interactions. The DECT-derived RSP maps could potentially be compromised by image noise levels and the severity of artifacts when using physics-based mapping techniques. This work presents a noise-robust learning-based method to predict RSP maps from DECT for proton radiation therapy. The proposed method uses a residual attention cycle-consistent generative adversarial network to bring DECT-to-RSP mapping close to a 1-to-1 mapping by introducing an inverse RSP-to-DECT mapping. To evaluate the proposed method, we retrospectively investigated 20 head-and-neck cancer patients with DECT scans in proton radiation therapy simulation. Ground truth RSP values were assigned by calculation based on chemical compositions and acted as learning targets in the training process for DECT datasets; they were evaluated against results from the proposed method using a leave-one-out cross-validation strategy. The predicted RSP maps showed an average normalized mean square error of 2.83% across the whole body volume and an average mean error less than 3% in all volumes of interest. With additional simulated noise added in DECT datasets, the proposed method still maintained a comparable performance, while the physics-based stoichiometric method suffered degraded inaccuracy from increased noise level. The average differences from ground truth in dose volume histogram metrics for clinical target volumes were less than 0.2 Gy for D95% and Dmax with no statistical significance. Maximum difference in dose volume histogram metrics of organs at risk was around 1 Gy on average. These results strongly indicate the high accuracy of RSP maps predicted by our machine-learning–based method and show its potential feasibility for proton treatment planning and dose calculation.