Physics-Guided Radiotherapy Treatment Planning with Deep Learning

2025-06-23

AI Summary

In radiotherapy, frequent replanning due to anatomical changes is time-consuming and highly experience-dependent. To address this, we propose a physics-guided two-stage deep learning framework for automated volumetric modulated arc therapy (VMAT) replanning. The first stage jointly predicts multileaf collimator (MLC) sequences and monitor unit (MU) parameters; the second stage incorporates physics-informed supervision derived from 3D dose distributions, integrating radiophysical priors and enforcing dose-driven optimization via inverse dose feedback. Leveraging a synergistic 3D U-Net and UNETR architecture, the method was validated on 133 prostate cancer patients: mean error in PTV D95% was 0.42 ± 1.83 Gy, V95% deviation was −0.22 ± 1.87%, and organ-at-risk doses were significantly reduced—achieving clinically acceptable plan quality. The key innovation lies in embedding interpretable, physics-based constraints directly into the end-to-end learning pipeline, thereby balancing automation efficiency with clinical safety.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Low Level & Physics-based Vision

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Radiotherapy (RT) is a critical cancer treatment, with volumetric modulated arc therapy (VMAT) being a commonly used technique that enhances dose conformity by dynamically adjusting multileaf collimator (MLC) positions and monitor units (MU) throughout gantry rotation. Adaptive radiotherapy requires frequent modifications to treatment plans to account for anatomical variations, necessitating time-efficient solutions. Deep learning offers a promising solution to automate this process. To this end, we propose a two-stage, physics-guided deep learning pipeline for radiotherapy planning. In the first stage, our network is trained with direct supervision on treatment plan parameters, consisting of MLC and MU values. In the second stage, we incorporate an additional supervision signal derived from the predicted 3D dose distribution, integrating physics-based guidance into the training process. We train and evaluate our approach on 133 prostate cancer patients treated with a uniform 2-arc VMAT protocol delivering a dose of 62 Gy to the planning target volume (PTV). Our results demonstrate that the proposed approach, implemented using both 3D U-Net and UNETR architectures, consistently produces treatment plans that closely match clinical ground truths. Our method achieves a mean difference of D95% = 0.42 +/- 1.83 Gy and V95% = -0.22 +/- 1.87% at the PTV while generating dose distributions that reduce radiation exposure to organs at risk. These findings highlight the potential of physics-guided deep learning in RT planning.

Problem

• Automating adaptive radiotherapy planning for anatomical variations
• Optimizing MLC and MU parameters using deep learning
• Reducing radiation exposure to organs at risk

Innovation

• Two-stage physics-guided deep learning pipeline
• 3D U-Net and UNETR architectures for planning
• Dose distribution supervision in training