Robust enhanced leaf model implementation for beam‐matched linear accelerators: Impact of physical leaf gap variations

BACKGROUND: Leaf-gap variability may influence MLC modeling when implementing the Enhanced Leaf Model (ELM) in Eclipse v18.0 for beam-matched linear accelerators (linacs) previously commissioned using legacy dosimetric leaf gap (DLG)-based models. PURPOSE: To characterize inter-machine variability in calibrated leaf gaps and associated ELM-derived MLC parameters across beam-matched linacs and to develop a practical strategy for robust multi-linac ELM implementation. METHODS: Five TrueBeam linacs equipped with Millennium 120 MLC were grouped into two clinical sites (Group A: three linacs; Group B: two linacs) sharing a matched beam model within each group. Variations in calibrated leaf-gap, measured DLG, and DLG (ELM) values were evaluated. Leaf-gap variation was harmonized by adjusting Gap value in the machine setting while maintaining a machine-averaged ELM across beam-matched linacs. ELM performance was evaluated using three MLC test plans: dynamic MLC (dMLC), narrow MLC (nMLC), and static MLC (sMLC). Following Gap adjustment, DLG was remeasured, and DLG (ELM) values were verified. Central doses of dMLC fields were measured using an ionization chamber and compared with ELM-calculated doses. Dose profiles were measured with radiochromic film, and were compared with calculated profiles before and after Gap adjustment. Profile agreement was further quantified using full width at half maximum (FWHM) and one-dimensional gamma analysis (2%/1 mm, 10% dose threshold). RESULTS: In Group A, ELM-derived MLC parameters showed minimal variation, with DLG (ELM) values agreeing within 0.02 cm among the linacs with Gap values set to zero for all linacs. Film measurements demonstrated excellent agreement between calculated and measured dose distributions for all MLC test fields, with nMLC gamma passing rates generally exceeding 98% and FWHM differences within 0.2 mm. In Group B, an initial 0.05 cm difference in Gap value resulted in substantial discrepancies in dMLC tests, including central dose deviations of -7.5% and +5.0% relative to the corresponding calculated doses and approximately 0.05 cm difference in the measured dMLC field width. Following Gap adjustment, ELM-derived parameters were converged and dosimetric agreement was improved markedly, with dMLC central dose differences reduced to within ± 1.1%. nMLC tests resulted in gamma pass rates exceeding 98%, and FWHM differences within 0.2 mm across all positions. CONCLUSIONS: The calibrated leaf-gap, represented by the Gap value in the machine setting, directly influences DLG (ELM), measured DLG and dMLC dose response. Dynamic sweeping-gap measurements provide a sensitive and efficient method for validating ELM performance and identifying residual inter-machine discrepancies. A hybrid strategy combining machine-averaged ELM parameters with harmonized leaf-gap settings provides a practical and robust approach for multi-linac ELM implementation.