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.

A Respiratory Motion Prediction Method Based on Improved Relevance Vector Machine.

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

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

Predicting respiratory motion for real-time tumour tracking in radiotherapy

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

respiTrack: Patient-specific real-time respiratory tumor motion prediction using magnetic tracking

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

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

TH-AB-303-02: an Investigation of Respiratory Signal Parameters for Multiple-Step Ahead Prediction of Surrogate Motion

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

Real-Time 2D MR Cine from Beam Eye’s View with Tumor-Volume Projection to Ensure Beam-to-Tumor Conformality for MR-Guided Radiotherapy of Lung Cancer

Respiratory Motion Estimation of Tumor Using Point Clouds of Skin Surface

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

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

A Respiratory Motion Prediction Method Based on LSTM-AE with Attention Mechanism for Spine Surgery

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

Volumetric prediction of breathing and slow drifting motion in the abdomen using radial MRI and multi-temporal resolution modeling

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

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

Modeling of artificial intelligence-based respiratory motion prediction in MRI-guided radiotherapy: a review

A Super-Learner Model for Tumor Motion Prediction and Management in Radiation Therapy: Development and Feasibility Evaluation.