BrachyUtils: A High-throughput and Modular Brachytherapy Treatment Planning System Built for Python Scripting

BrachyUtils is a modular Python framework designed to streamline reproducible research in high dose rate (HDR) brachytherapy. It addresses limitations in complex clinical workflows that constrain plan quality under time pressure, where automation tools exist but lack standardized benchmarking and integration. The framework integrates open-source and in-house tools via Docker for reproducibility, connecting key components: registration (OpenTPS, Plastimatch, SimpleElastix), dose calculation/analysis (RapidBrachyTG43, RapidBrachyMC), and optimization (Gurobi, AMPL, ORTools). It also manages data anonymization and file conversions. Benchmarking on the μ-RegPro prostate dataset and 13 in-house HDR prostate plans demonstrated its effectiveness. Contour-based registration outperformed image-based methods; SimpleElastix-BSpline achieved the top prostate Dice score (0.94 ± 0.02) and the lowest Hausdorff distance (4.87 mm ± 1.58 mm) for biopsy contours, while OpenTPS-rigid was fastest (0.89 s ± 0.01 s). Dosimetry comparisons showed that RapidBrachyTG43’s dose-to-water overestimated dose volume histogram (DVH) metrics by 3.3%–6.8% versus RapidBrachyMC’s dose-to-medium. Among solvers, Gurobi excelled with the shortest times (0.59 s ± 0.28 s) for linear/ quadratic penalties. Overall, BrachyUtils unifies data handling, registration, dosimetry, and optimization in a reproducible environment, enabling automated evaluations and accelerating treatment planning research. The rapidly evolving framework can be accessed at https://github.com/engerlab/brachyutils.git.