Département de Physique, Université de Montréal, Pavillon Roger-Gaudry, 2900 Boulevard Édouard-Montpetit, Montréal, Québec H3T 1J4, Canada. Author to whom any correspondence should be addressed.
Phys Med Biol. 2019 Jun 21;64(12):125024. doi: 10.1088/1361-6560/ab1e9d.
The purpose of this study is to evaluate the effect of an intravenous (IV) contrast agent on proton therapy dose calculation using dual-energy computed tomography (DECT). Two DECT methods are considered. The first one, [Formula: see text], attempts to accurately predict the proton stopping powers relative to water (SPR) of contrast enhanced (CE) DECT images, while the second generates a virtual non-contrast (VNC) volume that can be processed as a native non-contrast (NC) one. Both methods are compared against single-energy computed tomography (SECT). The accuracy of SPR predicted for different concentrations of IV contrast diluted in water is first evaluated using simulated data. Results then are validated in an experimental set-up comparing SPR predictions for both NC and CE images to measurements made with a multi-layer ionisation chamber (MLIC). Finally, the impact of IV contrast on dose calculation using both SECT and DECT is evaluated for one liver and one head and neck patient. Using simulated data, DECT is shown to be less sensitive to the presence of IV contrast than SECT, although the performance of the [Formula: see text] method is sensitive to the level of beam hardening considered. For different concentrations of IV contrast diluted in water, experimental MLIC measurement of SPR agrees with DECT predictions within 3% while SECT introduce errors above 20%. This error in the SPR value results in a range error of up to 3.2 mm (2.6%) for proton beams calculated on SECT CE patient images. The error is reduced below 1 mm using DECT with the [Formula: see text] and VNC methods. Globally, it is observed that the influence of IV contrast on proton therapy dose calculation is mitigated using DECT over SECT. In patient anatomies, the VNC approach provides the best agreement with the reference dose distribution.
本研究旨在评估在使用双能计算机断层扫描（DECT）时，静脉内（IV）造影剂对质子治疗剂量计算的影响。我们考虑了两种 DECT 方法。第一种方法，[公式：见文本]，试图准确预测对比增强（CE）DECT 图像中的质子相对水的阻止本领（SPR），而第二种方法生成一个可以作为原生非对比（NC）处理的虚拟非对比（VNC）体积。这两种方法均与单能计算机断层扫描（SECT）进行了比较。首先使用模拟数据评估不同浓度 IV 造影剂在水中稀释时 SPR 预测的准确性。然后在实验装置中验证结果，将 NC 和 CE 图像的 SPR 预测与使用多层电离室（MLIC）进行的测量进行比较。最后，评估使用 SECT 和 DECT 对一例肝脏和一例头颈部患者进行剂量计算的 IV 造影剂的影响。使用模拟数据表明，DECT 对 IV 造影剂的存在不如 SECT 敏感，尽管[公式：见文本]方法的性能对考虑的束硬化程度敏感。对于不同浓度的 IV 造影剂在水中稀释，MLIC 实验测量的 SPR 与 DECT 预测值在 3%以内一致，而 SECT 则引入了超过 20%的误差。SPR 值的误差导致基于 SECT CE 患者图像计算的质子束的范围误差高达 3.2 mm（2.6%）。使用 DECT 的[公式：见文本]和 VNC 方法可以将误差降低到 1mm 以下。总体而言，与 SECT 相比，DECT 可减轻 IV 造影剂对质子治疗剂量计算的影响。在患者解剖结构中，VNC 方法与参考剂量分布的吻合度最佳。
