A fully autonomous robotic ultrasound system for thyroid scanning.

Area of Science:

• Robotics
• Medical Imaging
• Artificial Intelligence

Background:

• Current thyroid ultrasound interpretation is subjective and demanding, relying on sonographer and radiologist expertise.
• The physical and cognitive load on clinicians during ultrasound procedures is significant.

Purpose of the Study:

• To develop and evaluate a fully autonomous robotic ultrasound system for thyroid scanning.
• To assess the system's ability to identify malignant thyroid nodules and assist in ACR TI-RADS classification.

Main Methods:

• The system integrates human skeleton point recognition, reinforcement learning, and force feedback for autonomous thyroid target localization.
• Bayesian optimization is employed for dynamic adjustment of the ultrasound probe's orientation.
• The system was tested on human participants for performance evaluation.

Main Results:

• The robotic system achieved high-quality ultrasound scans comparable to those performed by experienced clinicians.
• The system demonstrated capability in detecting thyroid nodules.
• The system has the potential to provide data for American College of Radiology Thyroid Imaging Reporting and Data System (ACR TI-RADS) calculations.

Conclusions:

• A fully autonomous robotic ultrasound system can perform high-quality thyroid scans independently.
• This robotic approach shows promise in improving diagnostic accuracy and reducing clinician workload in thyroid nodule detection.
• The system facilitates objective nodule characterization for standardized reporting systems like ACR TI-RADS.