A Trajectory Log–Based Approach to Fraction-Specific Quality Assurance
Abstract
Purpose
Modern linear accelerators maintain precise control of delivery parameters during the treatment of dynamic external beam radiotherapy treatment plans; vendors provide snapshots of these parameters using trajectory log files (TLF). This project aims to investigate the sensitivity of a commercial log file validation tool, allowing the implementation of fraction-specific quality assurance of these treatment plans.
Methods
TLFs for five patient treatment fractions were produced and collected from two TrueBeam Linear Accelerators, representing HD MLC (N=2) and Standard Millenium MLC (N=3). Initial comparisons between plans calculated in Eclipse V15.6 treatment planning system and by the commercial software Clearcalc V2.6.5 RadMonteCarlo (RMC) were followed by dose comparisons with an RMC dose calculation through control point positions provided in the trajectory log files. A custom-developed C# application, TrajectoryLog.net, modified the treatment TLFs to simulate common errors that might be experienced during clinical treatment; these include systematic leaf-position errors, mirroring those caused by mechanical wear (e.g. T-Nut wear, leading to MLC Backlash). The RMC tool calculated 3D gamma analyses with 2% and 2mm criteria within the plan’s target volume across all treated fraction TLFs and modified TLFs with a 0.5mm systematic leaf offsets across.
Results
RMC software was validated to detect simulated errors introduced within external beam radiotherapy treatment TLFs. RMC from the treatment plan control points and unmodified TLF RMC comparisons yielded a gamma pass rate ranging from [96.9%,100%] and [97.1%, 100.0%], respectively. Gamma pass rates for modified TLFs decreased by 0.70% on average, with all but one patient being negatively impacted by the modification.
Conclusion
Sensitivity testing of a commercial dose calculation engine shows daily auditing of TLFs can assist in detection of divergent plan delivery parameters. These results suggest a semi-automated workflow integrating RMC into regular fraction-specific QA can assist finding errors in machine deliverability.