Robot-assisted ultrasound probe calibration for image-guided interventions

Area of Science:

• Medical Imaging
• Robotics
• Surgical Navigation

Background:

• Trackable ultrasound probes enable augmented reality in ultrasound-guided procedures for precise lesion localization.
• Current manual calibration and registration methods are labor-intensive, time-consuming, and suboptimal.
• Accurate probe-to-image registration is crucial for effective fusion of ultrasound data and live video streams.

Purpose of the Study:

• To develop and validate a novel automated calibration approach for trackable ultrasound probes.
• To eliminate the need for manual manipulation in the calibration process.
• To improve the efficiency and accuracy of ultrasound probe registration.

Main Methods:

• A robotic manipulator executed stylus movements for automated probe positioning.
• A 6-degree-of-freedom electromagnetic tracker was integrated for real-time pose tracking.
• A feature detection algorithm identified stylus tip coordinates in ultrasound feeds for automatic correspondence selection.

Main Results:

• The automated system achieved a mean re-projection error of 0.38 mm, comparable to manual landmark selection (0.34 mm).
• Image plane reconstruction accuracy improved significantly with automatic segmentation (0.20 deg) versus manual (0.80 deg).

Conclusions:

• The proposed system enables fully automated calibration of trackable ultrasound probes with state-of-the-art accuracy.
• Streamlines recalibration after sterilization and simplifies factory calibration for mass-produced probes.
• Enhances the usability of trackable ultrasound probes in clinical and manufacturing settings.