Permanent breast seed implant (PBSI) brachytherapy is a novel technique for earlystage breast cancer. Computed tomography (CT) images are used for treatment planning and freehand 2D ultrasound for implant guidance. The multimodality imaging approach leads to discrepancies in target identification. To address this, a prototype 3D ultrasound (3DUS) system was recently developed for PBSI. In this study, we characterize the 3DUS system performance, establish QA baselines, and develop and test a method to register 3DUS images to CT images for PBSI planning. METHODS AND MATERIALS: 3DUS system performance was characterized by testing distance and volume measurement accuracy, and needle template alignment accuracy. 3DUS-CT registration was achieved through point-based registration using a 3D-printed model designed and constructed to provide visible landmarks on both images and tested on an in-house made gel breast phantom. RESULTS: The 3DUS system mean distance measurement accuracy was within 1% in axial, lateral, and elevational directions. A volumetric error of 3% was observed. The mean needle template alignment error was 1.0 AE 0.3 and 1.3 AE 0.5 mm. The mean 3DUS-CT registration error was within 3 mm when imaging at the breast centre or across all breast quadrants. CONCLUSIONS: This study provided baseline data to characterize the performance of a prototype 3DUS system for PBSI planning and developed and tested a method to obtain accurate 3DUS-CT image registration for PBSI planning. Future work will focus on system validation and characterization in a clinical context as well as the assessment of impact on treatment plans.

No abstract