Room: AAPM ePoster Library
Purpose: To establish an analytical model for multileaf collimator (MLC) geometric parameters, based only on information available in the literature, to produce an accurate Monte-Carlo (MC) model of the MLC.
Methods: The static part of the linear accelerator was replaced by a validated phase-space. An extensive literature review was performed to collect available geometric parameters and general constraints (maximum field, leaf number, collimator over-travel, source-isocenter and source-collimators distances) for a specific MLC, composed by leaves and perpendicular diaphragms with rounded ends. Analytical expressions for the radius, thicknesses and position of the collimators were determined as functions of the gathered information. A MLC MC model was constructed with the obtained values and used to simulate inter and intraleaf transmission in a water phantom. Simulations were compared to measurements acquired in a water tank with a micro-diamond detector. The virtual tongue-and-groove (vT&G), density and leaf thickness were set as adjustable parameters and iteratively fine-tuned until simulations and measurements agreed. For validation, 2D dose profiles in a water tank and 3D dose distributions in a commercial water-equivalent cylindrical phantom were simulated (with the tuned MC model), for square fields, and compared to measurements and simulations performed by a clinical treatment planning system (TPS), respectively, under the same conditions.
Results: MC simulations are especially sensitive to the leaf’s density, thickness and vT&G, which should be tuned for individual equipments. 2D simulated and measured profiles are in good agreement. Gamma passing rates over 95% (3%/3mm) are found for 3D dose distributions, simulated with MC and TPS using computed tomography data. Validation for clinically relevant fields on the cylindrical phantom is ongoing.
Conclusion: The parameters obtained by the analytical models can be used to produce a reliable equipment-specific geometry of MLCs for MC simulations without depending on vendors’ confidential information, which may not be easily accessible.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by the Brazilian National Council for Scientific and Technological Development (CNPq, Grant No. 201383/2015-2); the Saudi Ministry of Education; the Alexander von Humboldt Foundation; and the Munich-Centre for Advanced Photonics (MAP). The authors thank Prof. Dr. Guillaume Landry, Dr. Chiara Gianoli and Dr. Mark Podesta.