Deep learning based time series analysis for breathing phase prediction in phase-gated proton therapy

Clinical Snapshot

Key Findings

• P Value: Not reported in abstract; 'significantly outperformed' stated without explicit p-values in abstract

• Effect Size: Approximately 20% improvement in phase prediction accuracy for LSTM and TFT models over the commercial algorithm; timing deviations reduced across all four breathing regularity levels

• Primary Outcome: Phase prediction accuracy assessed by mean squared error (MSE), phase binning accuracy, and timing deviation for radiation on/off events; LSTM and TFT models significantly outperformed the commercial algorithm

• Nnt Or Sensitivity: Not applicable in traditional clinical sense; timing deviation reduction is the key operational metric — specific millisecond values not provided in abstract

• Confidence Interval: Not reported in abstract

Clinical Application

Clinical implementation would require integration of DL inference engines into existing proton therapy gating hardware and software pipelines. Real-time inference latency, system integration testing, regulatory approval (FDA 510(k) or equivalent), and prospective clinical validation are all prerequisites before routine deployment. Computational infrastructure requirements are not reported. Australia has a small but growing number of proton therapy facilities, including the Australian Bragg Centre for Proton Therapy and Research (Adelaide) and planned facilities in other states. Phase-gated proton therapy is relevant to these centres. The TGA would require regulatory approval for any DL-based gating software as a medical device software (SaMD) under the Software as a Medical Device framework. RANZCR and ACPSEM would be the relevant professional bodies for clinical implementation guidance. PBS does not currently fund proton therapy for most indications, limiting the immediate patient population, though this is evolving. The findings are technically relevant to Australian proton therapy physics practice but require local prospective validation before adoption. Adult patients undergoing phase-gated spot-scanned proton therapy for thoracic or abdominal malignancies where respiratory motion management is required; particularly relevant for patients with irregular breathing patterns

Abstract

Objective. Phase gating is a critical technique to mitigate tumor motion during radiotherapy, particularly in spot-scanned particle therapy where internal motion can interfere with dynamic spot scanning patterns and, simultaneously, introducing substantial range uncertainties. However, the current commercial state of the art in real-time phase prediction is challenged by patient-specific breathing variability as well as detection and delivery system latencies. This leads to suboptimal efficiency, mis-timed radiation delivery and requires frequent manual intervention. This study aims to improve phase prediction accuracy using deep learning (DL)-based time series forecasting to enable more accurate dose delivery.Approach. Retrospective breathing waveforms from 69 proton therapy patients, sampled at 30 Hz, were labeled with inspiratory peaks and assigned subjective regularity scores (four levels). DL models with various architectures were trained using waveform amplitude to predict current or future breathing phases. Model performance was evaluated using mean squared error, phase binning accuracy, and timing deviation for radiation on/off events. Bayesian optimization was used for hyperparameter tuning. Results were compared between models and against a commercial algorithm currently in clinical use.Main Results. The curated waveform dataset included 165 242 s for training, 24 057 s for validation, and 30 322 s for testing, with an additional 40 604 s from separate patients for extended validation. The long short-term memory and temporal fusion transformer models significantly outperformed the commercial algorithm, improving phase prediction accuracy by nearly 20% and reducing timing deviations across all regularity levels.Significance. DL-based time series forecasting may substantially improve breathing phase prediction accuracy over current clinically available methods, offering a more precise and reliable approach to phase-gated radiation delivery.

References

1. 1.Qian, J., Tang, X., Matysiak, W., Zhou, M., Stottler, Z. S., El-Sherif, O., Deisher, A. J., & Tryggestad, E. (2026). Deep learning based time series analysis for breathing phase prediction in phase-gated proton therapy. Physics in Medicine and Biology. https://doi.org/10.1088/1361-6560/ae7eec