Monitoring delivered dose today — predicting treatment needs tomorrow.
Radiation therapy is delivered over multiple fractions, and the dose actually received by a patient can differ from the original treatment plan as anatomy, positioning, and machine delivery conditions change. Our research combines independent Monte Carlo dose calculation, treatment delivery log data, and predictive modeling to reconstruct patient-specific delivered dose, monitor clinically meaningful variations over time, and support timely treatment adaptation.
Patient-specific Delivered-dose Reconstruction
We reconstruct the dose delivered during each treatment fraction using machine log data, including spot positions, monitor units, and beam parameters. Independent Monte Carlo calculation provides a physics-based estimate that does not rely solely on the treatment planning system and enables patient-specific assessment across photon, proton, and carbon-ion therapy.
Longitudinal Dose Monitoring
Fraction-by-fraction dose reconstruction makes it possible to track deviations that may not be visible in pre-treatment quality assurance alone. By evaluating delivered dose over the full treatment course, we investigate cumulative dose changes, systematic delivery trends, and inter-fractional variations that may affect treatment quality.
Prediction for Adaptive Radiation Therapy
We develop data-driven methods to predict clinically relevant dose deviations before they accumulate. By integrating prior-fraction delivery data, patient-specific dose metrics, and treatment trends, our goal is to identify when additional review or adaptive re-planning may be beneficial, including time-critical workflows in adaptive carbon-ion radiation therapy.
Monte Carlo–based Verification and Uncertainty Analysis
Monte Carlo simulation serves as an independent reference for dose reconstruction and prediction. We evaluate agreement using clinically relevant metrics, including gamma analysis, dose–volume histogram parameters, and patient-specific tolerance criteria, while investigating uncertainty from machine delivery, patient anatomy, and computational modeling.
Key Topics
- Independent Monte Carlo reconstruction of delivered dose
- Machine log–based fraction-by-fraction dose monitoring
- Longitudinal and cumulative patient-specific dose assessment
- Prediction of clinically relevant dose deviations
- Decision support for online and offline adaptive radiation therapy
- Photon, proton, and carbon-ion treatment applications
- Automation of dose monitoring and prediction workflows