New AI framework enables robotic ultrasound with precise probe control
Researchers developed an AI-powered system that tracks probe angles with 1.06° precision for better ultrasound imaging.
Researchers Xihan Ma and Haichong Zhang have developed an AI-powered robotic ultrasound framework that overcomes a critical limitation in traditional systems: fixed probe positioning. Most robotic ultrasound systems maintain probes perpendicular to the body surface, but this fails to capture diagnostic views requiring angled probes—such as in echocardiography. Their new framework, Omni-Directional Probe-Orientation Control, integrates real-time RGB-D perception with local surface modeling using multi-view point clouds.
The system fuses these inputs to create a quadratic surface estimate, enabling the probe to track arbitrary angles relative to the surface normal. In experiments, the framework achieved a mean angular tracking error of 1.06° ± 0.66° and successfully recovered non-normal tilt angles up to 44.39° ± 2.59°. The approach was validated on flat-surface tracking, phantom target-angle recovery, and in vivo cardiac imaging, demonstrating its potential for clinical applications.
- Achieves mean angular tracking error of 1.06° ± 0.66° for precise probe control
- Enables probe tilts up to 44.39° ± 2.59° relative to surface normal, far beyond traditional systems
- Validated on phantoms and in vivo experiments, including cardiac imaging
Why It Matters
Could revolutionize robotic ultrasound by enabling high-precision, non-normal probe angles for better diagnostics in cardiology and other specialties.