Robotic CBCT meets robotic ultrasound

Area of Science:

• Medical Imaging
• Robotics
• Interventional Radiology

Background:

• Current multi-modality imaging systems face dexterity and mobility limitations, hindering integration into clinical workflows and autonomous intervention.
• Advancing fully autonomous intervention systems requires novel approaches to imaging device integration and co-registration.

Purpose of the Study:

• To present a novel clinical setup with pre-calibrated robotic cone beam computed tomography (CBCT) and robotic ultrasound (US).
• To enable registration-free rigid registration and dynamic co-registration of robotic CBCT and US for multi-modal guided procedures.
• To facilitate new clinical applications by overcoming limitations of current imaging devices.

Main Methods:

• A one-time pre-calibration was performed between robotic CBCT and US systems.
• Vessel segmentation from B-mode US images was performed using SAM2, with Doppler signals as prompts, to highlight vasculature on 3D CBCT.
• Fused images were created by mapping US data (Doppler image or segmented vessel masks) onto CBCT, validated using a phantom with simulated lesions and moving flow.

Main Results:

• The mapping error between US and CBCT was an average deviation of 1.72 ± 0.62 mm.
• A user study confirmed the effectiveness of CBCT-US fusion for needle insertion guidance.
• Needle intervention performance improved by approximately 50% compared to conventional US-guided workflows, with enhanced time efficiency, accuracy, and success rates.

Conclusions:

• This work presents the first robotic dual-modality imaging system for guiding clinical applications.
• The developed system demonstrates significant performance improvements over traditional manual interventions.
• The findings pave the way for more advanced, autonomous image-guided procedures.