Abstract
The accurate placement of radioactive seeds in prostate brachytherapy is critical to the efficacy of the procedure. Current manual needle insertion methods face challenges, including reduced accuracy due to hand tremors, high dependence on surgeon expertise, and strain during lengthy procedures. Additionally, manual approaches often struggle to adapt to tissue heterogeneities, leading to unsatisfied outcomes. Autonomous needle placement is difficult due to varying tissue parameters. This paper presents an innovative dual-arm visual-servo robotics system for needle steering precision during prostate brachytherapy. The system employs two Franka Emika arms: one for needle insertion and the other for positioning the ultrasound probe. Based on the real-time position feedback, a fuzzy logic controller guides needle steering, and a camera system offers supplementary tracking and safety monitoring. In order to identify the needle tip's position within tissue, a novel image recognition method which is intuitive to the surgeon is proposed with the use of the ultrasound probe. It is in conjunction with the scanning and control mode of the dual-arm robotic arm to locate the position of the needle tip inside the tissue. The camera system is also unified in the same dual-arm robotic arm coordinate system to monitor the entire needle steering process. By addressing the limitations of manual techniques, including accuracy, efficiency, and adaptability to tissue variations, the proposed system reduces the skill barrier, workload, and potential trauma associated with brachytherapy procedures. Experimental validation on a phantom shows a final needle placement accuracy of 0.285 cm, demonstrating the system's potential to improve treatment outcomes through precise needle control.