Introduction
In the field of medical technology, two primary software ecosystems often operate in isolation: 3D Slicer, the standard for open-source medical image analysis, and the Robot Operating System (ROS), the foundation of modern robotics research. Integrating these platforms for image-guided interventions (IGT) has traditionally required custom, difficult-to-maintain "glue" code. Researchers from Queen’s University and Johns Hopkins University have addressed this gap with SlicerROS2, a high-performance C++ module designed to bridge these two ecosystems seamlessly.
The Integration Challenge
Developing image-guided robotic systems—such as robots performing biopsies under real-time MRI—requires a synchronized "common language" between imaging hardware and robotic actuators. Historically, research groups have relied on device-specific software, leading to incompatible protocols, high barriers to entry for new researchers, and a significant lack of code reproducibility across different robotic platforms.
SlicerROS2 aims to establish a standard integration scheme, allowing researchers to visualize and control medical robots directly within a sophisticated imaging environment without reinventing the underlying infrastructure.
System Architecture and Implementation
The module was redesigned from the ground up using rclcpp (the ROS 2 C++ API) to ensure high performance. By moving to a native C++ implementation, the authors achieved a 20x performance boost over previous Python-based iterations, reducing "spin" time from approximately 57ms to 2.4ms.
Key technical features include:
- MRML Integration: ROS 2 elements—including publishers, subscribers, parameters, and the
tf2transform tree—are mapped directly into 3D Slicer’s Medical Reality Modeling Language (MRML) data model. This allows ROS topics to appear as native data nodes within Slicer. - The Robot Node: This specialized node parses Unified Robot Description Format (URDF) files, enabling Slicer to automatically load and animate 3D robot models in real-time based on incoming ROS messages.
- Synchronized Execution: Utilizing a Qt timer to manage the ROS event loop at 50–100 Hz, the module ensures that robotic movements and image updates remain synchronized without compromising the user interface's responsiveness.
Experimental Evaluation
The system was evaluated using a variety of hardware and simulated environments, including the da Vinci Research Kit (dVRK), the Omni Bundle haptic device, and a "Smart Template" for prostate biopsy. Testing involved physical phantoms (ballistic gel with CT-visible markers) and Gazebo simulations for ultrasound data and kinematics.
Results and Performance
SlicerROS2 demonstrated robust performance metrics suitable for clinical research:
- Low Latency: Round-trip message latency between ROS and Slicer averaged between 8.5 ms and 11.9 ms, meeting the requirements for real-time surgical visualization.
- Virtual Fixtures: In breast surgery simulations, users experienced haptic feedback at tumor boundaries with an accuracy of 2.0–3.4 mm.
- Path Following: The system guided a da Vinci robot to trace paths around simulated tumors with an accuracy of 2.75 ± 0.65 mm.
- Stability: The 3D viewer maintained a steady 49 FPS while processing high-frequency robotic data streams.
Conclusion and Future Directions
SlicerROS2 represents a significant advancement in the accessibility of image-guided robotics research. By leveraging Slicer’s "observer" pattern, the system can trigger robotic actions based on imaging events, such as sending haptic "stop" commands when a tool reaches a segmented boundary.
While the module currently requires hardware with existing ROS interfaces and has yet to fully implement ROS 2 "Services" and "Actions," it successfully shortens the design-build-test cycle for custom medical devices. The project remains open-source, inviting community contributions to further expand its capabilities in the medical robotics domain.
Summary updated