Author: Muhammad Ramish Ashraf, Dixin Chen, Lloyd Emmanuel Kamole Ghomsi, Edward Elliot Graves, Billy W Loo, Stavros Melemenidis, Joseph B. Schulz, Lawrie Skinner 👨🔬
Affiliation: Department of Radiation Oncology, Department of Radiation Oncology, Stanford University School of Medicine 🌍
Purpose: Precise beam collimation is essential in preclinical radiotherapy studies to ensure accurate dose delivery. Manual design and fabrication of organ-specific collimators usually consist of multiple custom parts, including a high-Z mold and two-piece enclosing case. The design of these collimators is time-consuming and prone to error due to separate design adjustments, impacting reproducibility. A scalable automated workflow can streamline this process, improving efficiency and consistency.
Methods: A custom Python-based software package was developed to automate the parametric creation of collimator, providing geometries and alignment adjustments to all elements of the design via computer-aided-design (CAD). The platform enables scripting complex aperture designs or with defined size, shape, and positional offsets, producing 3D-models in STL format suitable for 3D-printing. In addition to other collimator components, generated molds were used to create the Cerrobend shielding collimator component. A 15×15mm2 partial-lung-collimator with a 7.5mm offset from the central-beam-axis and a 40x15mm2 whole-lung-collimator with no offset were fabricated and validated using film-based surface dose profiles under conventional (CONV) 11.9MeV and ultra-high dose rate (FLASH) 12.8MeV irradiation with a custom Varian TrueBeam-mounted-applicator. Full-width at half-maximum (FWHM) differences between CONV and FLASH were assessed for beam-shaping accuracy.
Results: High dimensional accuracy in 3D-printed components ensured precise collimator fabrication and additional parts. Film-based dose profiles confirmed alignment with intended designs, with FWHM differences of 0.3mm(X) and 0.4mm(Y) for partial lung and 0.2mm(X) and 0.4mm(Y) for whole lung between CONV and FLASH. The workflow greatly reduced design and production time due to trivial scalability.
Conclusion: The automated collimator creation platform increases efficiency and reproducibility in the fabrication of preclinical irradiation collimators. Beyond collimator fabrication, the STL export capability supports Monte Carlo simulations and treatment planning, enabling simultaneous inverse optimization of apertures. Future work will explore additional functionality for multi-angle collimator designs and compensator implementation to expand its applicability.