BACKGROUND AND PURPOSE Deformable image registration (DIR) facilitated dose reconstruction and accumulation can be applied to assess delivered dose and verify the validity of the treatment plan during treatment. This retrospective study used in silico deformations based on clinically observed anatomical changes as ground truth to investigate the uncertainty of reconstructed and accumulated dose in head-and-neck radiotherapy (HNRT). MATERIALS AND METHODS A planning CT (pCT), cone beam CT (CBCT) from week one of treatment and three later CBCTs were selected for 12 HNRT patients. These images were used to generate in silico reference CBCTs and deformation vector fields (DVFs) as ground truth with B-spline DIR. Inverse consistency (IC) of voxels was assessed by determining their net displacement after successive application of the forward and backward DVF. The reconstructed dose based on demons DIR was compared to the ground truth to assess the structure-specific uncertainties of this DIR algorithm for inverse consistent and inverse inconsistent voxels. RESULTS Overall, 98.5% of voxels were inverse consistent with the 95% level of confidence range for dose reconstruction of a single fraction equal to [-2.3%; +2.1%], [-10.2%; +15.2%] and [-9.5%; +12.5%] relative to their planned dose for target structures, critical organs at risk (OARs) and non-critical OARs, respectively. Inverse inconsistent voxels generally showed a higher level of uncertainty. CONCLUSION The uncertainty in accumulated dose using DIR can be accurately quantified and incorporated in dose-volume histograms (DVHs). This method can be used to prospectively assess the adequacy of target coverage during treatment in an objective manner.

中文翻译：

---

量化头颈部放射治疗中可变形图像配准后的剂量累积不确定性

背景和目的 可变形图像配准 (DIR) 促进剂量重建和累积可用于评估递送剂量并验证治疗期间治疗计划的有效性。这项回顾性研究用于基于临床观察到的解剖学变化的硅片变形作为基本事实，以研究头颈部放疗 (HNRT) 中重建和累积剂量的不确定性。材料和方法 为 12 名 HNRT 患者选择了治疗第一周的计划 CT (pCT)、锥束 CT (CBCT) 和 3 次以后的 CBCT。这些图像用于生成计算机参考 CBCT 和变形矢量场 (DVF) 作为具有 B 样条 DIR 的地面实况。在连续应用前向和后向 DVF 后，通过确定体素的净位移来评估体素的逆一致性 (IC)。将基于恶魔 DIR 的重建剂量与真实情况进行比较，以评估该 DIR 算法对于逆一致和逆不一致体素的结构特定不确定性。结果 总体而言，98.5% 的体素与等于 [-2.3%; +2.1%], [-10.2%; +15.2%] 和 [-9.5%; +12.5%] 相对于它们分别用于靶结构、危及危急器官 (OAR) 和非关键 OAR 的计划剂量。逆不一致体素通常表现出更高水平的不确定性。结论 使用 DIR 累积剂量的不确定性可以准确量化并纳入剂量体积直方图 (DVH)。该方法可用于以客观的方式前瞻性地评估治疗期间目标覆盖的充分性。