Abstract Background Consistent commissioning of novel radiation therapy systems remains a critical step to ensure safety and accuracy of radiation delivery. The initial installations of the Reflexion X1 linear accelerator (linac) and Treatment Planning System (TPS) allowed for development of a consistent method for beam model commissioning and convergence towards a consensus reference beam model. Purpose To define reference commissioning, quality assurance (QA), and quality control (QC) data for the RefleXion X1 linac and TPS. Methods Acquired beam measurements and TPS model data for the first seven installed X1 linear accelerators were compared. Each site's beam model was optimized based on (1) measurements of static beam geometry in liquid water and (2) treatment beam geometry in various phantoms using diodes, ion chambers, film, and array detectors. Treatment plans were generated in a consistent manner, and their measured doses were compared across all institutions. The observed differences in machine‐specific parameters as measured during QC, and the corresponding differences in beam models are reported. Evaluation of a candidate consensus model for all systems was also performed including cross comparison of measurements and different site‐specific beam models. Results Static beam data showed agreement for all X1s in percent depth dose, lateral beam profiles, and output factors, with inter‐linac differences generally < 2% in all metrics considered. Notable differences in longitudinal field widths and beam centering implies a consensus beam model is not ideal for the X1. Treatment planning and delivery also demonstrated larger differences which required unique beam model optimization for all treatment sites. Conclusions A systematic method for Reflexion planning and dosimetric commissioning was developed including a standardized set of treatment plans for model validation. Inter‐institution differences between measurements and beam models suggest a gold beam reference dataset may be achievable with improved QC at installation, but at this time individual, site‐specific beam models are necessary to encompass machine‐specific variations.
Watkins et al. (Wed,) studied this question.