ESTRO 2025 - Abstract Book

S314

Brachytherapy – Head & neck, skin, eye

ESTRO 2025

4357

Digital Poster Implementation of tissue heterogeneity corrections in 3D Monte Carlo library-based dose calculations for eye plaque brachytherapy planning Oleksii Semeniuk 1,2 , Mark J. Rivard 1,2 , Robert A. Weersink 3,4 1 Medical Physics, Brown University, Providence, USA. 2 Radiation Oncology, Rhode Island Hospital, Providence, USA. 3 Radiation Oncology, University of Toronto, Toronto, Canada. 4 Radiation Medicine Program, Princess Margaret Cancer Center, Toronto, Canada Purpose/Objective: Conventional low-dose-rate (LDR) eye plaque brachytherapy treatment planning is based on the TG-43 formalism, which overestimates dose to tumor and organs-at-risk (OAR) (1-6). Plaque Simulator and EyeDose (4, 6-9) assume a water medium, ignoring ocular heterogeneity. This approach underestimates tumor dose, while overestimating lens dose. Monte Carlo (MC) methods improve dosimetric accuracy, but are computationally intensive. To permit accurate and efficient dose calculations, we investigate a strategy of MC-based treatment planning that uses a dose kernel-scaling approach in which sets of MC simulations in water are pre-generated for different eye plaques. Material/Methods: Dose calculations of 16-mm COMS and 20-mm notched eye plaques with 125 I seeds were performed using the egs_brachy package of EGSnrc. To investigate dose distributions of standard plaques, the tumor was symmetrically located on the medial side of the eye. For the notched plaque, dose distributions were investigated with the tumor located posteriorly (abutting the optic nerve). Dose distributions with all active seeds (MC all ) were compared to those generated by the superposition of individual MC seed kernels into a library (MC library ) for generic and notched eye plaques. MC simulations were performed for two eye phantom models: a complete heterogeneous eye phantom and a homogeneous water-eye phantom using TG-43 calculations as a reference. In addition, a heterogeneity corrected distribution was obtained by scaling the water-eye phantom dose distribution by the quotient of energy absorption coefficients of corresponding eye structures and water. Results: Compared to complete phantom calculations, water phantom calculations underestimated tumor dose by ~17%, overestimated lens and sclera doses, while correctly predicting lens dose. The cDVH curves for heterogeneity corrected phantom essentially overlaid those for the complete phantom for both plaques (Fig. 1). Modulay shielding of the optic nerve in the notched plaque reduced optic nerve D20% from 42 Gy using TG-43 to 3 Gy using MC methods. The MC all simulations matched MC library within ~1% (Fig. 2). The largest differences between the heterogeneity-corrected vs. complete phantom doses were observed in D0% for tumor and sclera, but were within 5% of the prescription dose (Fig. 2).