Limitations of a fixed rise time model for beam delivery time prediction in dose-driven continuous scanning proton...

Dose-Driven Continuous Scanning (DDCS) proton therapy is designed to reduce treatment time. Its efficiency depends on accurate Beam Delivery Time (BDT) prediction. Manufacturer models use a simplified assumption of a single fixed beam current rise time (T_rise) and linear ramp-up after beam interruptions (break points).

This study validates the simplified T_rise, model, assesses its limitations, and identifies sources of inaccuracy. A numerical BDT model incorporating a fixed T_rise, was validated against machine log files from a Hitachi synchrotron. Experimental plans with varying energies (70.2-228.7 MeV) and currents (8-20 MU/s), as well as clinical patient plans were delivered. The modeled beam current profiles were compared against measured beam current profiles.

The manufacturer's default T_rise (3.0 ms) resulted in Mean Absolute Percentage Errors (MAPEs) of 20%-30% in BDT for patient plans. The optimal Trise was not a constant but was dependent on both beam energy and current, varying from 1.9 ms to 4.6 ms. Model error correlated with break point frequency; higher-energy layers exhibited lower BDT errors. Furthermore, complex beam dynamics unmodeled by the linear ramp-up assumption such as intra-spill current fluctuation and dips during spill changes were identified as important contributors to the overall BDT error.

A single, fixed T_rise value is insufficient for BDT prediction. The primary sources of inaccuracy are the dependency of the optimal T_rise on beam parameters and the impact of other unmodeled current dynamics. Accurate BDT prediction requires models that account for the entire temporal beam current profile.