Department of Radiation Oncology, University Hospital, LMU Munich, Munich 81377, Germany.
Fondazione Policlinico Universitario Agostino Gemelli IRCCS, Rome 00168, Italy.
Radiother Oncol. 2023 May;182:109555. doi: 10.1016/j.radonc.2023.109555. Epub 2023 Feb 21.
Magnetic resonance imaging guided radiotherapy (MRgRT) with deformable multileaf collimator (MLC) tracking would allow to tackle both rigid displacement and tumor deformation without prolonging treatment. However, the system latency must be accounted for by predicting future tumor contours in real-time. We compared the performance of three artificial intelligence (AI) algorithms based on long short-term memory (LSTM) modules for the prediction of 2D-contours 500ms into the future.
Models were trained (52 patients, 3.1h of motion), validated (18 patients, 0.6h) and tested (18 patients, 1.1h) with cine MRs from patients treated at one institution. Additionally, we used three patients (2.9h) treated at another institution as second testing set. We implemented 1) a classical LSTM network (LSTM-shift) predicting tumor centroid positions in superior-inferior and anterior-posterior direction which are used to shift the last observed tumor contour. The LSTM-shift model was optimized both in an offline and online fashion. We also implemented 2) a convolutional LSTM model (ConvLSTM) to directly predict future tumor contours and 3) a convolutional LSTM combined with spatial transformer layers (ConvLSTM-STL) to predict displacement fields used to warp the last tumor contour.
The online LSTM-shift model was found to perform slightly better than the offline LSTM-shift and significantly better than the ConvLSTM and ConvLSTM-STL. It achieved a 50% Hausdorff distance of 1.2mm and 1.0mm for the two testing sets, respectively. Larger motion ranges were found to lead to more substantial performance differences across the models.
LSTM networks predicting future centroids and shifting the last tumor contour are the most suitable for tumor contour prediction. The obtained accuracy would allow to reduce residual tracking errors during MRgRT with deformable MLC-tracking.
磁共振成像引导放疗（MRgRT）结合可变形多叶准直器（MLC）跟踪技术，可以在不延长治疗时间的情况下解决刚性位移和肿瘤变形的问题。然而，为了实时预测未来的肿瘤轮廓，必须考虑到系统的延迟。我们比较了三种基于长短期记忆（LSTM）模块的人工智能（AI）算法在预测 500ms 后 2D 轮廓的性能。
模型使用一家机构治疗的患者的电影磁共振图像进行训练（52 例患者，3.1 小时的运动数据）、验证（18 例患者，0.6 小时）和测试（18 例患者，1.1 小时）。此外，我们还使用了另一家机构治疗的 3 例患者（2.9 小时）作为第二个测试集。我们实现了 1）一种经典的 LSTM 网络（LSTM-shift），用于预测肿瘤中心点在上下和前后方向的位置，这些位置用于移动最后观察到的肿瘤轮廓。LSTM-shift 模型在离线和在线两种方式下进行了优化。我们还实现了 2）一种卷积 LSTM 模型（ConvLSTM），用于直接预测未来的肿瘤轮廓，以及 3）一种卷积 LSTM 与空间变换层相结合的模型（ConvLSTM-STL），用于预测用于变形最后肿瘤轮廓的位移场。
在线 LSTM-shift 模型的性能略优于离线 LSTM-shift，明显优于 ConvLSTM 和 ConvLSTM-STL。它在两个测试集中分别达到了 50%的 Hausdorff 距离为 1.2mm 和 1.0mm。发现较大的运动范围会导致模型之间的性能差异更大。
预测未来中心点并移动最后肿瘤轮廓的 LSTM 网络是最适合肿瘤轮廓预测的。所获得的准确性将允许在具有可变形 MLC 跟踪的 MRgRT 中减少残留的跟踪误差。
