Purpose: To investigate the potential of using low‐spatial resolution (LR) correlated‐sampling Monte Carlo (CMC) to implement rapid high‐resolution (HR) dose computation in heterogeneous media for low‐energy brachytherapy‐ seed planning. Methods: An in‐house MC engine (PTRAN) was used to simulate a prostate patient case using 78 I‐125 Model 6711 seeds. CMC, which scores the dose difference, DD, between highly correlated histories in homogeneous and inhomogeneous geometries, was previously shown to improve efficiency by factors of 5–10. A CMCLR DD solution on a LR 2×2×2 mm grid was interpolated onto a HR 1×1×1 mm grid, on which the homogeneous water TG‐43 solution was computed. This summation of the CMCLR and HR TG‐43 solutions, CMCLR‐HR, was evaluated against the conventional HR Monte Carlo heterogeneous dose simulation (MCHR). Both voxel‐by‐voxel differences between CMCLR‐HR and MCLR and impact on prostate CTV dose‐volume histograms (DVH) were quantified for a non‐water prostate tissue assignment and for voxel‐to‐ voxel tissue variations typical of a post‐implant CT. Results: The CMCLR‐ HR and MCHR DVHs agreed within 2% for both prostate assignments. 57.6%, 79.7, and 93.9% of the voxels had percentage errors relative to MCHR within 1%, 3%, and 5%, respectively, for uniform CTV media compared to 29.7%, 45.3%, and 59.5% voxel frequencies using post‐implant voxel assignments. Regions with errors exceeding 3% in the uniform medium case corresponded to high dose‐gradients contrasted with a spatially random distribution of errors for the non‐uniform prostate. Conclusions: CMCLR‐HR is accurate within 3% for regions with uniform‐like tissues and small dose‐gradients. CMCLR‐HR duplicates DVHs within 2% suggesting application that CMCLR‐HR is would be useful in Monte Carlo brachytherapy plan optimization.
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging