Abstract:Background Stereotactic body radiotherapy of thoracic and abdominal tumors has to account for respiratory intrafractional tumor motion. Commonly, an external breathing signal is continuously acquired that serves as a surrogate of the tumor motion and forms the basis of strategies like breathing‐guided imaging and gated dose delivery. However, due to inherent system latencies, there exists a temporal lag between the acquired respiratory signal and the system response. Respiratory signal prediction models aim to compensate for the time delays and to improve imaging and dose delivery. Purpose The present study explores and compares six state‐of‐the‐art machine and deep learning‐based prediction models, focusing on real‐time and real‐world applicability. All models and data are provided as open source and data to ensure reproducibility of the results and foster reuse. Methods The study was based on 2502 breathing signals (ttotal≈90 h) acquired during clinical routine, split into independent training (50%), validation (20%), and test sets (30%). Input signal values were sampled from noisy signals, and the target signal values were selected from corresponding denoised signals. A standard linear prediction model (Linear), two state‐of‐the‐art models in general univariate signal prediction (Dlinear, Xgboost), and three deep learning models (Lstm, Trans‐Enc, Trans‐TSF) were chosen. The prediction performance was evaluated for three different prediction horizons (480, 680, and 920 ms). Moreover, the robustness of the different models when applied to atypical, that is, out‐of‐distribution (OOD) signals, was analyzed. Results The Lstm model achieved the lowest normalized root mean square error for all prediction horizons. The prediction errors only slightly increased for longer horizons. However, a substantial spread of the error values across the test signals was observed. Compared to typical, that is, in‐distribution test signals, the prediction accuracy of all models decreased when applied to OOD signals. The more complex deep learning models Lstm and Trans‐Enc showed the least performance loss, while the performance of simpler models like Linear dropped the most. Except for Trans‐Enc, inference times for the different models allowed for real‐time application. Conclusion The application of the Lstm model achieved the lowest prediction errors. Simpler prediction filters suffer from limited signal history access, resulting in a drop in performance for OOD signals.

Fast and Accurate Online Sequential Learning of Respiratory Motion with Random Convolution Nodes for Radiotherapy Applications.

Respiratory motion forecasting with online learning of recurrent neural networks for safety enhancement in externally guided radiotherapy

Respiratory Motion Prediction with Empirical Mode Decomposition-Based Random Vector Functional Link

Ensemble framework based real-time respiratory motion prediction for adaptive radiotherapy applications

Prediction of the motion of chest internal points using a recurrent neural network trained with real-time recurrent learning for latency compensation in lung cancer radiotherapy

Prediction of the position of external markers using a recurrent neural network trained with unbiased online recurrent optimization for safe lung cancer radiotherapy

Predicting respiratory motion for real-time tumour tracking in radiotherapy

Predicting respiratory motion using a novel patient specific dual deep recurrent neural networks

Rapid Prediction of Respiratory Motion Based on Bidirectional Gated Recurrent Unit Network

Respiratory Prediction Based on Multi-Scale Temporal Convolutional Network for Tracking Thoracic Tumor Movement

Towards real-time respiratory motion prediction based on long short-term memory neural networks

SU‐FF‐J‐113: Real Time Tracking of Respiratory Tumor Motion Based on External Respiratory Output

UT Transform Based Tumor Respiratory Motion Estimation and Prediction for Radiosurgery Robot.

SU‐E‐J‐191: Motion Prediction Using Extreme Learning Machine in Image Guided Radiotherapy

Future frame prediction in chest cine MR imaging using the PCA respiratory motion model and dynamically trained recurrent neural networks

Online Prediction for Respiratory Movement Compensation: a Patient-Specific Gating Control for MRI-guided Radiotherapy.

Prediction-driven respiratory motion atlas formation for 4D image-guided radiation therapy in lung

Benchmarking machine learning‐based real‐time respiratory signal predictors in 4D SBRT

Improving respiratory signal prediction with a deep neural network and simple changes to the input and output data format

R‐MFE‐TCN: A correlation prediction model between body surface and tumor during respiratory movement

LGEANet: LSTM-global temporal convolution-external attention network for respiratory motion prediction