Abstract
PURPOSE: This study quantifies isocenter and off-axis geometric uncertainties using MultiMet Winston-Lutz (WL) tests to optimize gross tumour volume (GTV)-to-PTV (planning target volume) margin for single isocenter multiple target (SIMT) stereotactic body radiotherapy (SBRT) across three linear accelerators. METHODS AND MATERIALS: Geometric inaccuracies were quantified for Trilogy (HD-MLC, 6 MV SRS), TrueBeam (Millennium MLC), and TrueBeam STx (HD-MLC, 6-degrees-of-freedom (6DoF) couch) using a Sun Nuclear MultiMet-WL cube containing six tungsten carbide markers arranged along the superior-inferior axis. The electronic portal imaging device (EPID) images acquired at four cardinal gantry angles with varied collimator/couch rotations were analyzed using MultiMet-WL software (v2.1) to measure 3D (Δ) displacements for all the LINACs. The required GTV-to-PTV margins were calculated using a modified van Herk formula (2.5Σ+1.64σ), incorporating measured 3D displacements for isocenter and off-axis targets. RESULTS: The TrueBeam STx (HD-MLC/6DoF) demonstrated superior geometric accuracy, maintaining ≤0.5 mm isocenter precision and ≤0.59 mm off-axis targeting (3-7 cm). The Trilogy exceeded TG-142 tolerances (1.06 ± 0.59 to 1.09 ± 0.57 mm) at all targets, requiring 4 mm uniform margins, while the TrueBeam (MMLC) showed optimal variations (isocenter: 0.68 ± 0.34 mm; superior off-axis: 0.74 ± 0.36 mm). Both TrueBeam platforms achieved sub-millimeter accuracy but demonstrated direction dependency for off-axis targets, requiring 2-3 mm anisotropic margins. Notably, isotropic margins introduced up to 11% delineation errors for off-axis targets due to these machine-specific geometric variations, highlighting the imperative for platform-specific margin protocols in SIMT SBRT. CONCLUSION: This study demonstrates that routine analysis of MultiMet WL testing is an essential tool for the spatial accuracy of the LINAC to establish machine-specific PTV margin expansion in SIMT-SBRT, particularly for targets where rotational errors dominate.