Monte Carlo simulations of geometric deformation of Harrison–Anderson–Mick applicators used in intraoperative radiation therapy

Background The Harrison–Anderson–Mick (HAM) applicator is a high‐dose‐rate intraoperative radiotherapy (HDR‐IORT) applicator used to position Ir‐192 brachytherapy sources over surgically‐accessed tumor volumes or post‐resection tumor beds. Because of the lack of a 3D imaging system, dwell times are optimized pre‐surgery by a TG‐43–based treatment planning system (TPS) that assumes a perfectly flat applicator surface surrounded by an infinite water phantom. These assumed conditions are disparate from typical treatment conditions, especially in the pelvic regions, which often involve uneven patient surfaces and superficial irradiations with little to no backscatter material. Purpose Develop and validate a Monte Carlo (MC) model of HDR‐IORT treatment with a HAM applicator and use this validated model to quantify inaccuracies in the dose calculations due to the simplified conditions assumed in the planning process. Methods The Oncentra Brachy TPS was used to optimize dwell times and calculate the delivered relative dose distribution for a 6‐catheter, 8‐cm HAM applicator with 0.5‐cm dwell steps and 1.0‐cm catheter spacing. A TOPAS (v.3.9) MC model of this treatment was then developed and validated against the TPS dose calculations. Once validated, the TOPAS applicator model was modified to calculate the difference in doses due to changes in backscatter conditions, the incorporation of the actual applicator materials, and curvature of the applicator. Dose distributions were characterized using dose to the prescription point, percent depth doses, and two‐dimensional isodose curves. Results Validation between the MC and TPS calculations was successful within 1.5% over a depth of 5.0 cm in water. Negligible dose differences were calculated when assuming the applicator is made completely of water versus modeling all the individual applicator components. Conversely, semi‐infinite phantom geometry had more than 5% loss in dose to the prescription point due to the absence of backscatter. Most severely, curvature radii in the range of 5.4 cm (shallow curvature) to 0.9 cm (steep curvature) had dose differences of 5%–25%, regardless of whether the curvature was along the catheter length or transverse to it. Conclusions This study quantified the changes in dose due to material and geometric differences that are currently not accounted for by the TPS. While not accounting for the material of the applicator contributes to negligibly, the presence of backscattering material can contribute up to 5%. The radius of the bending of the applicator was found to potentially have the largest impact on the dosimetry of the central plane, with deviations up to 25%.