Three-Dimensional Characterization of Isocenter Accuracy on Halcyon Linear Accelerator As a Function of Off-Axis Distance and Mlc Layer
Abstract
Purpose
To characterize the three-dimensional (3D) isocenter behavior of the Halcyon linear accelerator as a function of off-axis distance and multileaf collimator (MLC) layer, and to evaluate the need for spatially resolved isocenter metrics beyond a single-value specification.
Methods
Winston–Lutz images were acquired for 5 ball bearings (BBs) embedded in a phantom positioned at 0, 3.6, 5.4, 7.7, and 10 cm from isocenter. For each BB, the radiation field was independently shaped by the distal and proximal MLC layers and eight gantry, and four collimator angles were sampled at each location. Isocenter displacement vectors were reconstructed for gantry, collimator, and combined gantry–collimator rotations. Mean isocenter offset, root-mean-square (RMS) deviation, and maximum 3D deviation radius were calculated. Three-dimensional displacements were further decomposed into gun–target (beam-axis) and lateral (in-plane and cross-plane) components.
Results
Across all distances and MLC layers, gantry rotation resulted in larger 3D envelope and RMS radii than collimator rotation (0.79 vs 0.26 mm; 0.58 vs 0.19 mm). For combined gantry–collimator motion, the 3D isocenter offset increased with distance from isocenter, from 0.7mm at central-axis to approximately 1.1–1.3 mm at 8–10 cm off-axis. Directional decomposition revealed anisotropic walkout that increased with distance from isocenter: the gun-target displacement dominated the off-axis walkout approaching 0.5mm at 10cm, followed by the in-plane displacement reaching 0.30–0.35 mm by 8–10 cm while the cross-plane motion remained minimal (mean ≤0.05 mm; max ≤0.08 mm). Distal MLC layer consistently produced larger deviations than proximal, increasing the 3D envelope by approximately 0.1–0.2 mm across all off-axis distances studied.
Conclusion
Halcyon off-axis isocenter walkout is gantry-dominated, MLC-layer dependent, off-axis distance-dependent, and strongly anisotropic. These findings show that single-value isocenter tolerances are insufficient to characterize clinically relevant performance and spatially resolved metrics could potentially assess the off-axis dose-cloud fidelity.