Yin Yunpeng, Zhang Weisha, Zou Lian, Liu Xiangxiang, Yu Luxin, Wang Ming
College of Nuclear Technology and Automation Engineering, Chengdu University of Technology, Chengdu, 610059, China.
Cancer Center, Sichuan Provincial People's Hospital, University of Electronic Science and Technology of China, Chengdu, 610072, China.
BMC Cancer. 2025 Feb 18;25(1):289. doi: 10.1186/s12885-025-13669-0.
Treatment planning systems (TPS) often exclude immobilization devices from optimization and calculation, potentially leading to inaccurate dose estimates. This study employed deep learning methods to automatically segment 3D-printed head and neck immobilization devices and evaluate their dosimetric impact in head and neck VMAT.
Computed tomography (CT) positioning images from 49 patients were used to train the Mask2Former model to segment 3D-printed headrests and MFIFs. Based on the results, four body structure sets were generated for each patient to evaluate the impact on dose distribution in volumetric modulated arc therapy (VMAT) plans: S (without immobilization devices), S (with MFIFs), S (with 3D-printed headrests), and S (with both). VMAT plans (P, P, P, and P) were created for each structure set. Dose-volume histogram (DVH) data and dose distribution of the four plans were compared to assess the impact of the 3D-printed headrests and MFIFs on target and normal tissue doses. Gafchromic EBT3 film measurements were used for patient-specific verification to validate dose calculation accuracy.
The Mask2Former model achieved a mean average precision (mAP) of 0.898 and 0.895, with a Dice index of 0.956 and 0.939 for the 3D-printed headrest on the validation and test sets, respectively. For the MFIF, the Dice index was 0.980 and 0.981 on the validation and test sets, respectively. Compared to P, P reduced the V for PGTVnx, PGTVnd, PGTVrpn, PTV1, and PTV2 by 5.99%, 6.51%, 5.93%, 2.24%, and 1.86%, respectively(P ≤ 0.004). P reduced the same targets by 1.78%, 2.56%, 1.75%, 1.16%, and 1.48%(P < 0.001), with a 31.3% increase in skin dose (P < 0.001). P reduced the V by 9.15%, 10.18%, 9.16%, 3.36%, and 3.28% (P < 0.001), respectively, while increasing the skin dose by 31.6% (P < 0.001). EBT3 film measurements showed that the P dose distribution was more aligned with actual measurements, achieving a mean gamma pass rate of 92.14% under the 3%/3 mm criteria.
This study highlights the potential of Mask2Former in 3D-printed headrest and MFIF segmentation automation, providing a novel approach to enhance personalized radiation therapy plan accuracy. The attenuation effects of 3D-printed headrests and MFIFs reduce V and D for PTVs in head and neck cancer patients, while the buildup effects of 3D-printed headrests increases the skin dose (31.3%). Challenges such as segmentation inaccuracies for small targets and artifacts from metal fasteners in MFIFs highlight the need for model optimization and validation on larger, more diverse datasets.
治疗计划系统（TPS）在优化和计算过程中常常将固定装置排除在外，这可能导致剂量估计不准确。本研究采用深度学习方法自动分割3D打印的头颈部固定装置，并评估其在头颈部容积调强弧形放疗（VMAT）中的剂量学影响。
使用49例患者的计算机断层扫描（CT）定位图像来训练Mask2Former模型，以分割3D打印的头枕和多叶独立准直器（MFIF）。基于这些结果，为每位患者生成四组身体结构，以评估其对容积调强弧形放疗（VMAT）计划中剂量分布的影响：S（无固定装置）、S（有MFIF）、S（有3D打印头枕）和S（两者都有）。为每组结构创建VMAT计划（P、P、P和P）。比较四个计划的剂量体积直方图（DVH）数据和剂量分布，以评估3D打印头枕和MFIF对靶区和正常组织剂量的影响。使用Gafchromic EBT3胶片测量进行患者特异性验证，以验证剂量计算的准确性。
Mask2Former模型在验证集和测试集上对3D打印头枕的平均平均精度（mAP）分别为0.898和0.895，Dice指数分别为0.956和0.939。对于MFIF，在验证集和测试集上的Dice指数分别为0.980和0.981。与P相比，P使高危临床靶区（PGTVnx）、大体肿瘤靶区（PGTVnd）、阳性淋巴结区（PGTVrpn）、计划靶区1（PTV1）和计划靶区2（PTV2）的V分别降低了5.99%、6.51%、5.93%、2.24%和1.86%（P≤0.004）。P使相同靶区的V分别降低了1.78%、2.56%、1.75%、1.16%和1.48%（P<0.001），皮肤剂量增加了31.3%（P<0.001）。P使V分别降低了9.15%、10.18%、9.16%、3.36%和3.28%（P<0.001），同时皮肤剂量增加了31.6%（P<0.001）。EBT3胶片测量表明，P的剂量分布与实际测量结果更吻合，在3%/3毫米标准下的平均伽马通过率为92.14%。
本研究突出了Mask2Former在3D打印头枕和MFIF分割自动化方面的潜力，为提高个性化放射治疗计划的准确性提供了一种新方法。3D打印头枕和MFIF的衰减效应降低了头颈癌患者计划靶区的V和D，而3D打印头枕的建成效应增加了皮肤剂量（31.3%）。小靶区分割不准确和MFIF中金属紧固件产生的伪影等挑战凸显了在更大、更多样化的数据集中进行模型优化和验证的必要性。
