Mapping Rounded Leaf Tip Inputs In Eclipse V18 to Restore Tunable Plan Dosimetric Agreement across an Ensemble of Linacs
Abstract
Purpose
To enable flexible, self-consistent tuning of Enhanced Leaf Model (ELM) in Eclipse V18 photon algorithms by deriving the requisite sweeping gap measurement required to replicate the classical Dosimetric Leaf Gap (DLG) used in pre-V18 models, thereby establishing configurable plan dose agreement across an ensemble of treatment machines.
Methods
Measured DLG is defined as the x-intercept of plotting transmission-corrected sweeping gap measurements. Therefore, a linear mapping establishes the requisite uncorrected sweeping gap inputs needed as ELM inputs. For example M4mm=4·A+b+Mclosed·(1-4/120), where A is the nominal slope, determined empirically for each energy to minimize error between calculated and measured sweeping gaps, and b=A·DLG. Validated classical DLG parameters from RTadmin were mapped to sweeping gap inputs to configure AcurosXB18.1. This configuration was then used to compare calculated plan doses against production AXB15.6, Delta4 measurements, and AcurosXB18.1 configured using measured sweeping gaps on a TrueBeam linac.
Results
Certain combinations of energy and MLC require deviation from measured sweeping gap to maximize plan dosimetric agreement. 6MV, 10MV, and 10FFF on HDMLC using measured sweeping gap causes AcurosXB18.1 to underdose 1-3% on high modulation plans compared to AXB15.6 and Delta4. Using pre-V18 classical DLGs (representing tuned ensemble), AXB18.1 dose agreement improves to within 0.5%. Other energies and MLC types also benefit from ensemble tunability when splitting the difference between a spread of leaf gaps across matched machines. The derived slope A for each energy minimizes error between measured and calculated sweeping gaps. While the dominant first order dosimetric landscape is dictated by DLG, ELM adds differential higher order blurring, resulting in slightly less sharp dose distributions as expected for rounded leaf ends.
Conclusion
We restore tunability and concordance between classical DLG and V18’s new ELM formalism through a robust linear mapping which overcomes the limitation of relying solely on individual sweeping gap measurements.