用Transformer预测放疗射野强度，评估其在真实临床扰动下的稳定性。
Robustness of Transformer-Based Fluence Map Prediction Under Clinically Realistic Perturbations
- 分两阶段建模：先由解剖结构预测剂量，再生成射野强度
- 严重旋转和噪声下性能骤降，但分层注意力模型更稳定
- 单纯用SSIM评价不靠谱，需结合物理约束检验
基于学习的射野强度预测为调强放射治疗（IMRT）提供了快速替代迭代逆向计划的方法，但其在真实分布偏移下的鲁棒性仍不明确。本文研究了一个两阶段Transformer流程，将解剖结构（CT与轮廓）映射至剂量，再映射至射野束强度图。对比了分层、全局及混合注意力的Transformer骨干网络，采用物理信息损失以保证能量一致性。在前列腺IMRT数据集上评估了几何扰动、辐射噪声、训练数据减少及域偏移下的鲁棒性，并在公开数据集上进一步验证了剂量阶段的表现。结果表明，在中等扰动下性能平稳下降，但在严重旋转和噪声下出现急剧失效；分层Transformer（如SwinUNETR）的上四分位能量误差增长更慢，表现出更好鲁棒性。此外，仅使用SSIM无法捕捉临床相关的误差，凸显物理信息评估的必要性。
原文摘要 · Abstract (English)
Learning-based fluence map prediction offers a fast alternative to iterative inverse planning in intensity-modulated radiation therapy (IMRT), but its robustness under realistic distribution shifts remains unclear. We study a two-stage transformer pipeline that maps anatomy (CT and contours) to dose and then to beamlet fluence maps. We compare fluence-stage transformer backbones with hierarchical, global, and hybrid attention, trained with a physics-informed loss enforcing energy consistency. Robustness is evaluated under geometric perturbations, radiometric noise, reduced training data, and domain shifts using a prostate IMRT dataset, with additional evaluation of the dose stage on public datasets. Results show smooth degradation under moderate perturbations but sharp failures under severe rotations and noise. Hierarchical transformers (e.g., SwinUNETR) exhibit slower growth in upper-quartile energy error, indicating improved robustness. We further show that SSIM alone fails to capture clinically relevant errors, highlighting the need for physics-informed evaluation.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。