文献速递消化道肿瘤放疗放疗物理与技术

融入患者个体化先验临床知识提升直肠癌在线自适应放疗临床靶区自动分割的多中心泛化性验证

Incorporating patient-specific prior clinical knowledge to improve clinical target volume auto-segmentation generalisability for online adaptive radiotherapy of rectal cancer: A multicenter validation.
PubMed 文献 · Radiother Oncol · 2024-12-13

本研究纳入三个中心的中危或局部晚期直肠癌患者,将治疗前扫描的患者个体化直肠系膜临床靶区勾画与基于人群的靶区间变形变异结合生成患者个体化权重图,训练带先验(MRI+先验)和不带先验(仅MRI)的两个自动分割模型,并在两个外部数据集上验证。两个外部中心中MRI+先验模型在分割指标上均显著优于仅MRI模型(P值<0.01),与外部基线模型无显著差异。结论认为加入患者个体化权重图使模型对机构偏好更稳健,性能与本地从头训练的模型相当,适合推广至多中心。

要点

  • 将治疗前靶区勾画作为患者个体化先验可使直肠系膜临床靶区自动分割模型对机构偏好更稳健。
  • 两个外部中心中MRI+先验模型的分割指标均显著优于仅MRI模型(P值<0.01)。
  • MRI+先验模型与外部基线模型之间无显著差异。

核心结论

  • 加入患者个体化权重图使临床靶区分割模型对机构偏好更稳健,性能与本地从头训练的模型相当,适合推广至多中心。

研究设计与样本

  • 纳入来自三个中心的中危或局部晚期直肠癌患者。
  • 将治疗前扫描的患者个体化直肠系膜临床靶区勾画与基于人群的靶区间可能变形变异结合生成患者个体化权重图。
  • 用本院数据训练带先验(MRI+先验)和不带先验(仅MRI)的两个直肠系膜临床靶区自动分割模型,并将两模型应用于两个外部数据集。
  • 为一个外部中心从头训练不带先验的外部基线模型。
  • 采用DSC、表面Dice、95HD和MSD评价分割性能。

主要发现

  • 在两个外部中心,MRI+先验模型在分割指标上均显著优于仅MRI模型(P值<0.01)。
  • 外部基线模型与MRI+先验模型之间无显著差异。

关键数值与推荐

  • 在两个外部中心,MRI+先验模型对比仅MRI模型的分割指标差异P值<0.01。
  • 外部基线模型与MRI+先验模型之间无显著差异。

临床怎么用

  • 原文结论认为该融入患者个体化先验的方法适合推广至多中心,未给出具体临床操作推荐。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND & PURPOSE: Deep learning (DL) based auto-segmentation has shown to be beneficial for online adaptive radiotherapy (OART). However, auto-segmentation of clinical target volumes (CTV) is complex, as clinical interpretations are crucial in their definition. The resulting variation between clinicians and institutes hampers the generalizability of DL networks. In OART the CTV is delineated during treatment preparation which makes the clinician intent explicitly available during treatment. We studied whether multicenter generalisability improves when using this prior clinical knowledge, the pre-treatment delineation, as a patient-specific prior for DL models for online auto-segmentation of the mesorectal CTV. MATERIAL & METHODS: We included intermediate risk or locally advanced rectal cancer patients from three centers. Patient-specific weight maps were created by combining the patient-specific CTV delineation on the pre-treatment scan with population-based variation of likely inter-fraction mesorectal CTV deformations. We trained two models to auto-segment the mesorectal CTV on in-house data, one with (MRI + prior) and one without (MRI-only) priors. Both models were applied to two external datasets. An external baseline model was trained without priors from scratch for one external center. Performance was evaluated on the DSC, surface Dice, 95HD and MSD. RESULTS: For both external centers, the MRI + prior model outperformed the MRI-only model significantly on the segmentation metrics (p-values < 0.01). There was no significant difference between the external baseline model and the MRI + prior model. CONCLUSION: Adding patient-specific weight maps makes the CTV segmentation model more robust to institutional preferences. Performance was comparable to a model trained locally from scratch. This makes this approach suitable for generalization to multiple centers.

来源:Radiother Oncol,查看原文。版权归原作者所有。

引用

  1. Ferreira Silvério N, van den Wollenberg W, Betgen A, et al. Incorporating patient-specific prior clinical knowledge to improve clinical target volume auto-segmentation generalisability for online adaptive radiotherapy of rectal cancer: A multicenter validation. Radiother Oncol. 2024. PMID: 39675574. doi:10.1016/j.radonc.2024.110667

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。