2963 - Automated Plan, Transfer, and Delivery Verification for Clinical Proton Therapy

Purpose/Objective(s): Pencil beam spot scanning (PBS) proton therapy involves a large number of control points with high-dimensional multi-leaf collimator (MLC) and spot parameters, making manual review or sampling-based verification insufficient to fully assess plan quality, data transfer integrity, and treatment delivery accuracy. We developed an automated framework to perform comprehensive, control-point–level verification of proton treatment plans, plan transfer, and delivered treatment records to support safe and efficient clinical PBS proton therapy.

Materials/Methods: Five proton PBS treatment plans were generated in a Treatment Planning System (TPS) and transferred to a Record-and-Verify (R&V) system via a Spot Map Conversion (SMC) process that subsequently drives proton treatment delivery. Patient, plan, and delivery data from both TPS and R&V systems were retrieved directly using Structured Query Language (SQL) queries through Open Database Connectivity (ODBC) connections implemented in a custom application. Extracted planning parameters included patient demographics, planning images and contours, prescription details, optimization settings, beam geometry, isocenter, energy layer and MLC configurations, snout positions, spot sizes, positions, weights, and plan status. Planning parameters were validated for deliverability and plan quality. Corresponding parameters from the R&V system were organized and compared against TPS values to assess plan transfer integrity and SMC accuracy. Treatment delivery records written back to the R&V system were compared with intended parameters to verify delivery accuracy.

Results: TPS parameter retrieval and plan quality verification were completed within 20 seconds per plan. Following SMC, a typical plan with approximately 20 energy layers, each containing tens of MLC points and approximately 10 spots, was converted into several hundred control points in the R&V system, resulting in more than 10,000 total MLC and spot data points. All MLC positions, spot positions, and spot weights agreed within numerical precision between TPS and R&V systems. Plan transfer integrity verification required 30–40 seconds per plan, and treatment delivery verification required approximately 80 seconds for a typical five-fraction treatment.

Conclusion: We developed an automated, database-driven verification framework for proton therapy that enables comprehensive plan quality assessment, plan transfer validation, and treatment delivery verification at the control-point level. By addressing the scale and complexity of PBS proton delivery, this approach improves workflow efficiency, mitigates plan transfer and delivery risks, and strengthens patient safety for routine clinical proton therapy. The framework was implemented during system commissioning to support the safe transition to clinical operation as patient treatments begin.