CRANE: A highly dexterous needle placement robot for evaluation of interventional radiology procedures

Dimitri A. Schreiber, Hanpeng Jiang, Guosong Li, Julie Yu, Zhaowei Yu, Renjie Zhu, Alexander M. Norbash, Michael C. YiparXiv 2019 · OpenReview

Abstract

Interventional Radiology (IR) enables earlier diagnosis and less invasive treatment of numerous ailments. Here we present our ongoing development of CRANE: CT RoboticArm and Needle Emplacer, a robotic needle positioning system for CT guided procedures. The robot has 8 active Degrees-of-Freedom (DoF) and a novel infinite travel needle insertion mechanism. The control system is distributed using the RobotOperating System (ROS) across a low latency network that interconnects a real-time low-jitter controller, with a desktop computer which hosts the User Interface (UI) and high-level control. This platform can serve to evaluate limitations in the current procedures and to prototype potential solutions to these challenges in-situ.

Introduction

Interventional Radiology (IR) relies heavily on CT scans to guide needles into precise locations for biopsies or pain management. However, the current manual process is cumbersome: physicians must repeatedly move the patient in and out of the CT bore to check the needle's position, leading to long procedure times and increased radiation exposure for both the patient and the clinician. Furthermore, hitting small targets (under 1cm) or reaching deep masses is technically demanding. Existing robotic solutions often struggle with the cramped space inside a CT scanner or lack the dexterity needed to navigate complex anatomical structures like the spine.

CRANE (CT RoboticArm and Needle Emplacer) is a new 8-degree-of-freedom (DoF) robotic system designed to bridge this gap. By combining a dexterous, cable-driven arm with a novel "infinite travel" needle driver, the system aims to reduce patient radiation exposure and improve the accuracy of high-precision needle placements during CT-guided procedures.

System Design and Methods

The researchers developed a hybrid robotic architecture consisting of two main sections designed to maximize both accuracy and reach within the confined space of a CT scanner.

The CRANE robotic system, highlighting the exo-bore backend (A) and the dexterous intra-bore arm (B).
  • The Exo-Bore Backend: A 4-DoF positioning stage (X-Y-Z Cartesian stage plus a revolute axis) that sits outside the scanner. This provides a stable, high-precision base for the rest of the system.
  • The Intra-Bore Arm: A 4-DoF serial-link arm that extends into the scanner. To keep the profile slim and avoid bulky motors near the patient, this arm is cable-driven, with the motors located back on the exo-bore structure.
  • Infinite Travel Needle Driver: A novel feature utilizing a clutching mechanism with Shape Memory Alloy (SMA) actuators. These actuators allow the robot to "grip and reset" the needle, effectively providing infinite insertion depth despite a physically short insertion stage.
  • Distributed Control System: The software is built on the Robot Operating System (ROS). It splits tasks between a "Master PC" (handling the user interface, 3D Slicer integration, and path planning) and a "Slave" controller (a System-on-Chip FPGA/ARM board) that runs real-time motor control at 1kHz to ensure smooth, safe motion.

Clinical Workflow

The system is designed to integrate seamlessly into existing medical workflows. It plugs into software like 3D Slicer, allowing physicians to plan needle paths directly on the patient's CT data. The proposed operational sequence ensures the doctor remains in control of the procedure while benefiting from robotic precision.

Flowchart detailing the integration of the CRANE system into standard interventional radiology procedures.

Key Insights and Results

The CRANE architecture demonstrates several key technical achievements:

  • 8-DoF Dexterity: The system successfully coordinates 8 axes of motion to provide a wide range of approach angles, significantly boosting dexterity in tight spaces compared to previous iterations.
  • Safety and Transparency: The use of lower gear ratio motors allows for "torque-sensing," enabling the robot to detect unexpected obstacles mechanically.
  • Miniaturization: Using SMA for the needle clutch achieves high gripping force in a tiny form factor, essential for operation near the patient's body inside the scanner bore.
  • High-Speed Control: The low-level controller achieves a 1kHz update rate, ensuring high-fidelity tracking of physician commands.

Limitations and Future Work

While the hardware and control systems are functional, this represents ongoing development. Current limitations include a lack of clinical validation on phantom models or synthetic tissues and the absence of specific quantitative accuracy metrics (e.g., sub-millimeter precision).

Future work will focus on clinical evaluations to assess performance during actual IR procedures. The team intends to perform a critical analysis of the system's accuracy and reliability to demonstrate its potential to outperform manual needle placement.

Summary updated

Abstract

Interventional Radiology (IR) enables earlier diagnosis and less invasive treatment of numerous ailments. Here we present our ongoing development of CRANE: CT RoboticArm and Needle Emplacer, a robotic needle positioning system for CT guided procedures. The robot has 8 active Degrees-of-Freedom (DoF) and a novel infinite travel needle insertion mechanism. The control system is distributed using the RobotOperating System (ROS) across a low latency network that interconnects a real-time low-jitter controller, with a desktop computer which hosts the User Interface (UI) and high-level control. This platform can serve to evaluate limitations in the current procedures and to prototype potential solutions to these challenges in-situ.
BibTeX
@misc{schreiber2019crane,
  author = {Dimitri A. Schreiber and Hanpeng Jiang and Guosong Li and Julie Yu and Zhaowei Yu and Renjie Zhu and Alexander M. Norbash and Michael C. Yip},
  title = {{CRANE: A highly dexterous needle placement robot for evaluation of interventional radiology procedures}},
  year = {2019},
  eprint = {1910.03063},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1910.03063},
}
Figure