Is model-based dose calculation based on cone-beam computed tomography suitable for adaptive treatment planning in brachytherapy?

Model-based dose calculation considering tissue compositions is increasingly being investigated in brachytherapy. The aim of this study was to assess the suitability of modern cone-beam computed tomography (CBCT) imaging compared to conventional computed tomography (CT) scans for this purpose.

By means of a phantom study, we evaluated the CT numbers and electron densities measured using a modern CBCT device as well as a conventional CT scanner for various materials. Based on this, we compared dose calculations (using the TG-43 formalism as well as model-based collapsed cone calculations assuming uniform materials [ACE_uniform] and considering CT numbers [ACE_CT#]) on planning CTs and control CBCTs for patients with cervical and breast cancer as well as phantom-simulated skin cancer cases. Assessing dosimetric deviations between the planning CTs and control CBCTs acquired during the treatment course served to estimate interfractional implant variations.

The comparison of ACE_uniform–ACE_CT# deviations between planning CTs and control CBCTs revealed no statistically significant difference for almost all examined dose parameters. Dosimetric deviations between model-based dose calculations and TG-43 were partly significant but of small magnitude (< 10 cGy per fraction). Interfractional dosimetric variations were substantially larger than the dosimetric differences found between the various dose calculation procedures.

Model-based dose calculation based on modern CBCT imaging was suitable. However, the found differences between these calculations and the TG-43 formalism should be investigated in dose–outcome analyses. The observed interfractional dosimetric variations revealed the importance of performing treatment quality assurance.