An Automated Raystation Plan-Check System for Error Reduction and Patient Safety
Abstract
Purpose
Manual plan checking is time-consuming and prone to variability across reviewers. To improve consistency and reliability in the plan-check workflow, we developed a fully automated RayStation plan-check system that performs more than 30 checks across eight key domains. The system standardizes the review process, reduces human error, and supports safer and more consistent patient treatment.
Methods
Using the RayStation (2024A) Python API, we implemented a modular QA engine in which all checks use deterministic, predefined pass/fail criteria. Each check evaluates TPS-accessible parameters against predefined thresholds and reports both the measured value and the corresponding acceptance criterion. Key checks include: CT & Imaging: simulation date, slice thickness, slice count, localization, and correct CT dataset selection (e.g., average, breath-hold) Prescription: fractional dose, number of fractions, prescription type, and dose-volume requirements Machine & Dose Grid: correct beam model and appropriate dose-grid assignment Approval: plan approval and reviewer identification Structure Checks: presence of required structures and compliance with expected types (e.g., External, Target, Couch) Beam & Geometry: energies, collimator angles, couch-angle settings, beam-name and arc-direction consistency Plan Quality: max-dose location, target coverage, normalization-range checks, modulation limits and clinical goals Isocenter & Setup: unintended isocenter shifts, multiple-isocenter discrepancies, clearance concerns, isocenter-to-PTV thresholds, and setup-beam alignment A standardized HTML report is generated with color-coded pass/fail indicators, displayed values, and acceptance criteria, supporting rapid review and QA traceability.
Results
The system was applied to over 200 clinical VMAT, SBRT, and 3D plans. It reliably identified predefined failing conditions, including beam-geometry mismatches, isocenter deviations, incorrect prescription inputs, unintended dose-grid selection, and excessive modulation. Average runtime was under 3 seconds per plan.
Conclusion
This automated plan-check system reduces reviewer variability, detects critical errors early, and enhances patient safety through deterministic, standardized QA checks. It offers a scalable, modernized approach to radiation therapy plan verification.