Identifying visible tissue in intraoperative ultrasound: a method and application

Area of Science:

• Medical Imaging
• Surgical Technology
• Robotics

Background:

• Intraoperative ultrasound scanning is a complex visuotactile task requiring precise probe manipulation and tissue characterization.
• Current methods face challenges in accurately assessing probe-tissue contact and identifying critical features like acoustic shadows.

Purpose of the Study:

• To develop and evaluate an advanced algorithm for analyzing probe-tissue contact during intraoperative ultrasound.
• To enhance the detection of acoustic shadows and improve confidence mapping of tissue perception.

Main Methods:

• An iterative filtering and topological method was employed to identify visible tissue and analyze probe-tissue interactions.
• Novel frameworks were developed for evaluating acoustic shadow classification and probe-tissue contact confidence.

Main Results:

• The proposed method demonstrated superior acoustic shadow classification on in vivo data (F-beta score of 0.864) compared to existing techniques.
• It achieved significantly better probe-tissue contact confidence estimation on phantom data (NRMSE of 0.168).
• The algorithm's robustness to various perturbations and its potential for robotic scan guidance were confirmed.

Conclusions:

• The developed algorithm shows significant potential for clinical application in ultrasound.
• It can enhance clinical training and enable greater automation in robotic ultrasound procedures.