A Semi-automated Robotic System for Percutaneous Interventions

Purpose: A robotic assistive device is developed for needle-based percutaneous interventions. The aim is a hybrid system using both manual and actuated robotic operation in order to obtain a device that has a large workspace but can still fit in the gantry opening of a CT scanner. This will enable physicians to perform precise and time-efficient CT-guided percutaneous interventions. The concept of the mechanics and software of the device is presented in this work.

Methods: The approach is a semi-automated robotic assistive device, which combines manual and robotic positioning to reduce the number and size of necessary motors. The system consists of a manual rough positioning unit, a robotic fine positioning unit and an optical needle tracking unit. The resulting system has eight degrees of freedom, of which four are manual, which comprise encoders to monitor the position of each axis. The remaining four axes are actuated axes for fine positioning of the needle. Cameras are attached to the mechanical structure for 3D tracking of the needle pose. The software is based on open-source software, mainly ROS2 as robotic middleware, Moveit2 for trajectory calculation and 3D Slicer for needle path planning.

Results: The communication between the components was successfully tested with a clinical CT scanner. In a first experiment, four needle insertions were planned and the deviation of the actual needle path from the planned path was measured. The mean deviation from the needle path to the target point was 21.9 mm, which is mainly caused both by translational deviation (15.4 mm) and angular deviation (6.8°) of the needle holder. The optical tracking system was able to detect the needle position with a mean deviation of 3.9 mm.

Conclusion: The first validation of the system was successful which proves that the proposed concept for both the hardware and software is feasible. In a next step, an automatic position correction based on the optical tracking system will be integrated, which is expected to significantly improve the system accuracy.