El‐Sherif, Omar

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.

AbstractBackgroundA proton therapy system with 190° gantries uses robotic couch rotations to change the treatment beam laterality. Couch rotations are typically validated clinically with post‐rotation radiographic imaging.AimsThis study assesses the specificity and sensitivity of a commercial 3D surface imaging system, AlignRT (Vision RT, London UK) for validating couch rotations.Materials & MethodsIn clinical operation, a reference surface image of the patient is acquired after radiographic setup with couch at 270°, perpendicular to the gantry axis of rotation. The couch is then rotated ±90° to a typical treatment angle, and AlignRT reports a 3D displacement vector. Patient motion, changes in patient surface, non‐coincidence between AlignRT and couch isocenter, and mechanical couch run‐out all contribute to the 3D vector magnitude. To assess AlignRT sensitivity in detecting couch run‐out, volunteers were positioned orthogonal to the proton gantry and reference surface images were captured without x‐ray localization. Subjects were repeatedly rotated ±90⁰ to typical treatment angles and displacement vectors were recorded. Additionally, measurements were performed in which intentional translations of 2, 4, 6, and 8 mm were combined with the intended isocentric rotations. Data sets were collected using a phantom; subjects with a thoracic isocenter and no immobilization; and subjects with a cranial isocenter and thermoplastic immobilization. A total of 300 rotations were measured.ResultsDuring isocentric rotations, the mean AlignRT displacement vectors for the phantom, immobilized, and non‐immobilized volunteers were 0.1 ± 0.1 mm, 0.8 ± 0.1 mm, and 1.1 ± 0.2 mm respectively. 95% of the AlignRT measurements for the immobilized and non‐immobilized subjects were within 1 mm and 2 mm of the actual displacement respectively.DiscussionAfter characterizing the accuracy using phantoms and volunteers, we have shown that a three‐pod surface imaging system can be used to identify gross non‐isocentric patient rotations. Significant positional deviations, either due to improper couch rotation or patient motion, should be followed by radiographic imaging and repositioning.ConculsionAlignRT can be used to verify patient positioning following couch rotations that are applied after the initial x‐ray guided patient setup. Using a three‐pod AlignRt system, positional deviations exceeding 4 mm were flagged with sensitivity and specificity of 90% and 100% respectively.