Ultrasound-Guided Robotic Assistance in Low-Dose-Rate Breast Brachytherapy

Thomas Lang, Kirill Makhachev, Mahdi Tavakoli

Keywords: Brachytherapy, Breast Cancer, Biomedical, Robotics

Abstract: Low-dose-rate (LDR) brachytherapy is an effective, localized radiation treatment for cancer. The procedure can be difficult to accurately place the seeds in highly deformable tissue like breast tissue. To address these challenges, this paper presents a robotic assistance system designed for breast brachytherapy to improve a clinician's situational awareness and reduce mental strain. The system uses a robotic arm for automated ultrasound (US) scanning and a U-Net convolutional neural network to generate a precise 3D model of the tumor's seroma. During the procedure, the system tracks the needle with an electromagnetic sensor and provides the clinician with a real-time view of both the needle and a 3D-reconstructed virtual environment. A feasibility study found that seed placement was significantly more accurate when using the system's combined views (Mode 4), which also resulted in the lowest reported mental load and frustration among subjects. These findings suggest that the system could improve both the qualitative and quantitative outcomes of LDR breast brachytherapy.

Submission Number: 11