Kumarasiri Akila, Liu Chang, Siddiqui Farzan, Zhong Hualiang, Chetty Indrin J, Kim Jinkoo
Department of Radiation Oncology, Henry Ford Health System, Detroit, Michigan.
Department of Radiation Oncology, Henry Ford Health System, Detroit, Michigan.
Pract Radiat Oncol. 2015 Jul-Aug;5(4):e317-25. doi: 10.1016/j.prro.2015.01.008. Epub 2015 Mar 5.
To estimate the delivered dose to targets and organs at risk for head and neck radiation therapy by accumulating the dose using 3 deformable image registration (DIR) algorithms.
Five head and neck patients, who had daily cone beam computed tomography (CT) images taken during the course of treatment, were retrospectively studied. To overcome the small field of view limitations and Hounsfield unit uncertainties of cone beam CT, a planning CT was deformably registered and resampled onto each cone beam CT image with a 4.2-cm uniform field of view expansion. The "dose of the day" was calculated on these resampled CT images, warped, and accumulated to the planning CT using 3 different DIR algorithms. Dosimetric indices for targets and organs at risk were determined from dose-volume histograms and compared with corresponding planned quantities.
There were no significant differences among the cumulative dose-volume histograms estimated by the 3 DIR algorithms. The cumulative mean dose deviation was less than 2% from the corresponding plan dose in general for the planning and gross tumor volumes. However, the parotid gland mean dose showed a large variation, with a maximum 33% deviation. This was due in part to considerable patient weight loss during the first 3 weeks of treatment. The corresponding target and organ mean dose deviations ± standard deviation, estimated as an average of the 3 DIR algorithms, were 1.0 ± 1.6% for planning target volumes, 1.6 ± 2.3% for gross tumor volumes, 7.3 ± 9.6% for left parotid, 10.3 ± 11.9% for right parotid, and 3.3 ± 4.7% for mucosa. The target coverage deviation (dose to 95% of the volume) was -2.8 ± 1.8% for planning target volumes and 0.2 ± 2.8% for gross tumor volumes. The deviation of the maximum dose to the spinal cord was 2.1 ± 2.4%.
Results of this study indicate that the estimated target dose generally remains within 2% of the intended dose for plans with a 5-mm planning margin. More frequent plan adaptation might be beneficial to avoid unintended excessive dose to parotid glands.
通过使用3种可变形图像配准（DIR）算法累积剂量，估算头颈部放射治疗中靶区和危及器官的实际照射剂量。
回顾性研究了5名头颈部患者，这些患者在治疗过程中每天都进行锥形束计算机断层扫描（CT）成像。为了克服锥形束CT视野小的局限性和亨氏单位的不确定性，将计划CT进行可变形配准，并重新采样到每个视野均匀扩展4.2厘米的锥形束CT图像上。在这些重新采样的CT图像上计算“当日剂量”，进行变形处理，并使用3种不同的DIR算法累积到计划CT上。根据剂量体积直方图确定靶区和危及器官的剂量学指标，并与相应的计划量进行比较。
3种DIR算法估算的累积剂量体积直方图之间无显著差异。计划靶体积和大体肿瘤体积的累积平均剂量偏差一般比相应的计划剂量低2%以内。然而，腮腺平均剂量变化较大，最大偏差为33%。部分原因是治疗的前3周患者体重明显减轻。以3种DIR算法的平均值估算，相应的靶区和器官平均剂量偏差±标准差分别为：计划靶体积1.0±1.6%，大体肿瘤体积1.6±2.3%，左侧腮腺7.3±9.6%，右侧腮腺10.3±11.9%，黏膜3.3±4.7%。计划靶体积的靶区覆盖偏差（体积的95%所接受的剂量）为-2.8±1.8%，大体肿瘤体积为0.2±2.8%。脊髓最大剂量偏差为2.1±2.4%。
本研究结果表明，对于有5毫米计划边界的计划，估算的靶区剂量一般保持在预期剂量的2%以内。更频繁地调整计划可能有助于避免腮腺意外接受过量剂量。
