Meixner Eva, Glogauer Benjamin, Klüter Sebastian, Wagner Friedrich, Neugebauer David, Hoeltgen Line, Dinges Lisa A, Harrabi Semi, Liermann Jakob, Vinsensia Maria, Weykamp Fabian, Hoegen-Saßmannshausen Philipp, Debus Jürgen, Hörner-Rieber Juliane
Department of Radiation Oncology, Heidelberg University Hospital, Im Neuenheimer Feld 400, 69120 Heidelberg, Germany.
Heidelberg Institute of Radiation Oncology (HIRO), 69120 Heidelberg, Germany.
Clin Transl Radiat Oncol. 2024 Sep 11;49:100855. doi: 10.1016/j.ctro.2024.100855. eCollection 2024 Nov.
Target volume delineation is routinely performed in postoperative radiotherapy (RT) for breast cancer patients, but it is a time-consuming process. The aim of the present study was to validate the quality, clinical usability and institutional-specific implementation of different auto-segmentation tools into clinical routine.
Three different commercially available, artificial intelligence-, ESTRO-guideline-based segmentation models (M1-3) were applied to fifty consecutive reference patients who received postoperative local RT including regional nodal irradiation for breast cancer for the delineation of clinical target volumes: the residual breast, implant or chestwall, axilla levels 1 and 2, the infra- and supraclavicular regions, the interpectoral and internal mammary nodes. Objective evaluation metrics of the created structures were conducted with the Dice similarity index (DICE) and the Hausdorff distance, and a manual evaluation of usability.
The resulting geometries of the segmentation models were compared to the reference volumes for each patient and required no or only minor corrections in 72 % (M1), 64 % (M2) and 78 % (M3) of the cases. The median DICE and Hausdorff values for the resulting planning target volumes were 0.87-0.88 and 2.96-3.55, respectively. Clinical usability was significantly correlated with the DICE index, with calculated cut-off values used to define no or minor adjustments of 0.82-0.86. Right or left sided target and breathing method (deep inspiration breath hold vs. free breathing) did not impact the quality of the resulting structures.
Artificial intelligence-based auto-segmentation programs showed high-quality accuracy and provided standardization and efficient support for guideline-based target volume contouring as a precondition for fully automated workflows in radiotherapy treatment planning.
乳腺癌患者术后放疗（RT）中，靶区勾画是一项常规操作，但这是一个耗时的过程。本研究的目的是验证不同自动分割工具在临床常规应用中的质量、临床可用性和机构特定实施情况。
将三种基于人工智能且符合欧洲放射肿瘤学会（ESTRO）指南的不同商用分割模型（M1 - 3）应用于50例连续的接受术后局部放疗（包括乳腺癌区域淋巴结照射）的参考患者，用于勾画临床靶区：残留乳腺、植入物或胸壁、腋窝1级和2级、锁骨上下区域、胸肌间和内乳淋巴结。使用骰子相似性指数（DICE）和豪斯多夫距离对生成结构进行客观评估，并对手动评估可用性。
将分割模型生成的几何形状与每位患者的参考体积进行比较，在72%（M1）、64%（M2）和78%（M3）的病例中无需或仅需进行微小校正。生成的计划靶区体积的DICE中位数和豪斯多夫值分别为0.87 - 0.88和2.96 - 3.55。临床可用性与DICE指数显著相关，计算得出的用于定义无需或进行微小调整的临界值为0.82 - 0.86。靶区的左右侧以及呼吸方式（深吸气屏气与自由呼吸）对生成结构的质量没有影响。
基于人工智能的自动分割程序显示出高质量的准确性，并为基于指南的靶区轮廓勾画提供了标准化和高效的支持，这是放疗治疗计划中完全自动化工作流程的前提条件。
