深度学习驱动的预测性治疗计划用于肺癌自适应放疗
本研究开发了序列分析深度学习算法Seq2Seq,基于计划CT预测肺癌根治性放疗期间肿瘤和食管每周的几何演变,并用每周CBCT更新预测,将预测融入预测性治疗计划范式。Seq2Seq在60例患者数据集上训练和验证,预测与实际的DICE值肿瘤为0.71–0.83、食管为0.71–0.75,平均Hausdorff距离在2 mm以内;与仅基于初始肿瘤和食管优化的计划相比,预测性周计划在维持60 Gy肿瘤覆盖的同时使食管平均剂量降低4.2 Gy。结论认为用Seq2Seq预测肿瘤和食管的纵向变化是可行的。
要点
- Seq2Seq基于计划CT预测肺癌放疗期间肿瘤和食管每周几何变化,并用每周CBCT更新预测。
- 预测与实际的DICE值:肿瘤0.71–0.83,食管0.71–0.75;平均Hausdorff距离在2 mm以内。
- 预测性周计划较仅基于初始肿瘤和食管的计划使食管平均剂量降低4.2 Gy,同时维持60 Gy肿瘤覆盖。
核心结论
- 用Seq2Seq预测肿瘤和食管的纵向变化是可行的,可能提高肺癌自适应放疗的效率与效果。
研究设计与样本
- 开发序贯分析深度学习算法Seq2Seq,使用包含60例患者的数据集进行训练和验证。
- Seq2Seq以计划CT上观察到的原发肿瘤和食管为起点,按周预测其放疗期间的几何演变,并用每周CBCT获取的新图像更新预测。
- Seq2Seq配备卷积长短期记忆网络分析纵向图像的空间-时间变化。
- 根据每周预测优化预测性计划并准备按周部署,以减轻在线每周重新计划的临床负担。
主要发现
- 预测与实际的每周肿瘤DICE为0.83 ± 0.10、0.79 ± 0.14、0.78 ± 0.12、0.77 ± 0.12、0.75 ± 0.12、0.71 ± 0.17。
- 预测与实际的每周食管DICE为0.72 ± 0.16、0.73 ± 0.11、0.75 ± 0.08、0.74 ± 0.09、0.72 ± 0.14、0.71 ± 0.14。
- 预测与实际的每周肿瘤和食管平均Hausdorff距离均在2 mm以内。
- 与仅使用初始肿瘤和食管优化的计划相比,预测性周计划使食管平均剂量降低4.2 Gy,同时维持60 Gy肿瘤覆盖。
- 食管平均剂量降低主要归因于放疗期间明显的肿瘤缩小。
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 AND PURPOSE: To develop a novel deep learning algorithm of sequential analysis, Seq2Seq, for predicting weekly anatomical changes of lung tumor and esophagus during definitive radiotherapy, incorporate the potential tumor shrinkage into a predictive treatment planning paradigm, and improve the therapeutic ratio. METHODS AND MATERIALS: Seq2Seq starts with the primary tumor and esophagus observed on the planning CT to predict their geometric evolution during radiotherapy on a weekly basis, and subsequently updates the predictions with new snapshots acquired via weekly CBCTs. Seq2Seq is equipped with convolutional long short term memory to analyze the spatial-temporal changes of longitudinal images, trained and validated using a dataset including sixty patients. Predictive plans were optimized according to each weekly prediction and made ready for weekly deployment to mitigate the clinical burden of online weekly replanning. RESULTS: Seq2Seq tracks structural changes well: DICE between predicted and actual weekly tumor and esophagus were (0.83 ± 0.10, 0.79 ± 0.14, 0.78 ± 0.12, 0.77 ± 0.12, 0.75 ± 0.12, 0.71 ± 0.17), and (0.72 ± 0.16, 0.73 ± 0.11, 0.75 ± 0.08, 0.74 ± 0.09, 0.72 ± 0.14, 0.71 ± 0.14), respectively, while the average Hausdorff distances were within 2 mm. Evaluating dose to the actual weekly tumor and esophagus, a 4.2 Gy reduction in esophagus mean dose while maintaining 60 Gy tumor coverage was achieved with the predictive weekly plans, compared to the plan optimized using the initial tumor and esophagus alone, primarily due to noticeable tumor shrinkage during radiotherapy. CONCLUSION: It is feasible to predict the longitudinal changes of tumor and esophagus with the Seq2Seq, which could lead to improving the efficiency and effectiveness of lung adaptive radiotherapy.
引用
- Lee D, Hu YC, Kuo L, et al. Deep learning driven predictive treatment planning for adaptive radiotherapy of lung cancer. Radiother Oncol. 2022. PMID: 35189155. doi:10.1016/j.radonc.2022.02.013
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。