Author: Michael Ashenafi, Nicholas Becerra Espinosa, Mario Ramos Gallardo, Matthew Pacella, Sean M. Tanny 👨🔬
Affiliation: Department of Radiation Oncology, University of Rochester 🌍
Purpose: The newest version of a large commercial treatment planning system (TPS) has introduced a new physical MLC model. We compare the impact of using an averaged dosimetric leaf gap (DLG) value versus machine-specific DLG values across four Varian TrueBeam linear accelerators using the updated MLC model. We quantify the extent of calculated dose changes within the expected DLG range.
Methods: Measurements were performed on four Varian TrueBeam linear accelerators following the vendors recommended DLG procedure for 6MV beams. DLG values were calculated using the enhanced leaf model (ELM) in Varian Eclipse (v18.0) TPS. Head-and-neck plans were recalculated using four approaches: (1) using a single machine’s measured DLG, (2) an average DLG from minimally varying measurements (DLG <.1mm variation, -0.0379cm), (3) an average including an outlier (DLG > .3mm variation, -0.0295), and (4, 5) the maximum vendor-recommended DLG range values (-0.1cm and 0.05cm).
Results: Target mean dose changes in scenarios 2 and 3 were <0.2% in comparison to the machine-specific DLG calculation. Mean dose to targets in scenarios 4 and 5 changed by ±3.5%. Depending on the prescription, this percent change can correspond as much as ±250cGy. All targets met sufficient coverage. Target coverage was unaffected in all scenarios. Max body dose differences were minimal (±13cGy) for scenarios 2 and 3 but reached ±225cGy for scenarios 4 and 5. Considering fractionation, these changes are not expected to have large clinical significance.
Conclusion: Based on measurements, it is not expected for DLG to vary dramatically between machines. The minimal variation in average DLGs vs. machine-specific suggest that using a general DLG is sufficient with negligible clinical consequence. The observed changes in dose at extreme cases were clinically insignificant. These results suggest that a universal DLG is reasonable within the scope of this work.