3037 - Comparison of Novel Automated and Conventional QA Methods for HDR Gynecological Brachytherapy Applicators

Purpose/Objective(s): Verification of source position is a critical component of annual quality assurance (QA) for high-dose-rate (HDR) gynecologic brachytherapy (GYN BT) applicators. Conventional workflows, film-based autoradiography, and CT imaging, can be labor-intensive, require careful manual handling, and are susceptible to setup variability and user-dependent errors. To address these challenges, we developed an automated workflow that integrates image registration and source position determination using applicator-specific 3D-printed phantoms. This study compares the new automated approach with a previously established manual measurement method to evaluate its accuracy, robustness, and potential to streamline clinical QA processes.

Materials/Methods: 31 Elekta HDR GYN applicators (Utrecht, Geneva, Venezia, and multichannel cylinder (MCC)) were evaluated. In the manual reference workflow, applicators were affixed to Gafchromic EBT-3 film and scanned after irradiation. In the automated workflow, custom 3D-printed phantoms were designed for each applicator type to securely immobilize the applicators with slits beneath them for fixed placement of EBT-3 film. Four X-SPOT radiographic markers were placed at the film corners to enable spatial registration. CT images were acquired with markers inserted into each applicator channel. The automated software registered the CT and scanned film images, calculating the positional differences between the irradiated source dwell positions and the markers. Source positions were quantified with 2D pixel-based measurements converted to actual distances. Mean displacement and range were calculated for each applicator type and compared between workflows to evaluate the accuracy and reproducibility of the automated approach.

Results: Across all applicators, mean source displacement was 0.32 ± 0.03 mm for the manual reference workflow and 0.33 ± 0.03 mm for the automated workflow. In the automated method, mean displacement by applicator type was: Utrecht 0.57 ± 0.05 mm (range: 0.33–0.9 mm), Geneva 0.41 ± 0.02 mm (range: 0.09–0.9 mm), Venezia 0.13 ± 0.01 mm (range: 0.02–0.27 mm), and MCC 0.38 ± 0.05 mm (range: 0.09–0.79 mm). All measurements showed submillimeter agreement, with no displacement exceeding 1 mm. Differences between the manual and automated workflows were small and not clinically significant.

Conclusion: The automated workflow demonstrated comparable submillimeter accuracy to the manual reference method while substantially reducing user interaction, setup variability, and potential for human error. By combining automated image registration with standardized phantom geometry, this approach provides a reproducible and scalable framework for HDR GYN applicator QA. Its implementation can streamline annual verification, enhance consistency across applicator types, and support broader adoption of quantitative, automation-driven QA practices, representing a meaningful advance in clinical BT workflow and safety.