Investigating Systematic Errors In Model-Based Dose Calculation Algorithms Using the TG‑186 Generic Applicator
Abstract
Purpose
Commercial model-based dose calculation algorithms (MBDCAs) show promise for advancing intensity-modulated brachytherapy (IMBT). However, transitioning from the TG‑43 protocol to MBDCAs requires benchmark studies against reference dosimetry. This work introduces a normalized-dose methodology to better characterize systematic errors in MBDCA implementations, using the TG‑186 generic applicator and two commercial algorithms as a test case.
Methods
TG‑186 test case 4 data were used with the 192Ir source centered in the applicator. Dose distributions from BrachyVision ACUROS (BVA) and Oncentra ACE were compared against Monte Carlo (MC). For each algorithm, three-dimensional dose matrices were normalized to a point 5 mm beyond the applicator surface along the source central axis, consistent with the reference point used for global dose-difference metrics in prior work. This normalization enabled voxelwise comparison of normalized isodose values as a coverage-based assessment. Conventional local and global dose-difference distributions were also reconstructed and compared with the proposed normalized-dose method.
Results
In unshielded regions, BVA isodose lines closely matched those from MC. In shielded regions, dose discrepancies of up to ~20% were observed near the distal and proximal ends of the shielding material, with the largest deviations (up to ~50%) at the apical tip and across the central lumen. ACE showed discrepancies of up to ~60% in both centrally shielded and some unshielded regions. When interpreted solely through local or global dose-difference maps, these discrepancies appeared unrealistically high or low, respectively, making their impact on coverage difficult to judge. The normalized-dose comparison provided a more intuitive visualization of where coverage could be compromised.
Conclusion
The normalized-dose distribution methodology offers a practical way to localize and interpret systematic dose discrepancies in MBDCAs for shielded applicators and complements conventional local and global dose-difference metrics during the transition from TG‑43 to model-based calculations.