Session Item

Proton therapy is rapidly growing worldwide due to the potential to deliver a highly conformal dose to the tumor while sparing healthy tissues. Our clinic recently acquired a MEVION S250i™ with HYPERSCAN™ pencil beam scanning (Figure 1a) that uses a range shifter to modulate the beam energy and has an Adaptive Aperture^TM (Multileaf Collimator) that generates a sharper lateral penumbra. An independent dose calculation method is desired due to the treatment complexities, which also requires treatment delivery verification (e.g. log file evaluation). This work describes an independent platform for Monte Carlo (MC) dose calculation and log files evaluation. A neuro case (15 fractions) with 3 beam directions using robust optimization on the CTV is shown as an example.

The treatment plan and log files were imported into an in-house developed software, A Medical Image-Based Graphical platfOrm (AMIGO, Figure 1b), for visualization and analysis. AMIGO has TPS-like features working as an interface for TOPAS (TOol for PArticle Simulation). All the relevant machine components (e.g. range shifter and adaptative aperture) were modeled using TOPAS 4D geometrical features. First, the log file information (e.g. range shifter position, spot position, monitor units (MU) per pulse, etc.) is compared to the treatment plan. Secondly, TOPAS dose calculations obtained using the treatment plan and log file parameters are compared.

Figure 1c shows the difference between log files (15 fractions) and the treatment plan for the position of each leaf of the dynamic aperture system. Deviations are below ± 0.7 mm. Spot positions have mean deviations of 0.0 ± 0.5 (1STD) and 0.1 ± 0.5 (1STD) for the inline and crossline directions, respectively. The monitor unit (MU) mean deviation is 0.00 ± 0.03 (1STD) for pulses, and 0.00 ± 0.01 (1STD) for spot positions (one spot can have multiple pulses). Total MU deviations range from 6.5 MU (≈0.3%) up to 17.5MU (≈0.8%). Figure 1d shows DVHs calculated using log files information for different fractions. Clinical metrics (e.g. D₉₀ and V₁₀₀) deviations are less than 1% for the CTV.

A continuous and consistent analysis of the log files provides valuable information about the machine condition (e.g. an increasing number of interlocks or systematic shifts in the measurements might indicate hardware failure in an early stage). Pre-treatment QA based on log files can reduce the workload in the clinic reducing the number of measurements. In addition, dose recalculations using an independent calculation method based on log files verifies the machine parameters during dose delivery and TPS calculations.  The workflow can be automated so log file based verification won’t have a negative impact in the workflow.