Adaptive median binary patterns for fully automatic nerves tracking in ultrasound images

Area of Science:

• Medical Imaging
• Anesthesiology
• Computer Vision

Background:

• Ultrasound-Guided Regional Anesthesia (UGRA) is crucial for targeted local anesthetic delivery.
• Operator skill is essential for UGRA, with nerve tracking in ultrasound images being a significant challenge due to image noise and artifacts.

Purpose of the Study:

• To develop a robust and fully automatic nerve tracking technique for ultrasound images.
• To enhance the accuracy and reliability of nerve tracking in Ultrasound-Guided Regional Anesthesia.

Main Methods:

• Introduced Adaptive Median Binary Pattern (AMBP) as a texture feature for tracking algorithms.
• Incorporated Kalman filter for prediction and correction to improve accuracy and handle target disappearance.
• Evaluated tracking algorithms including particle filter, mean-shift, and Kanade-Lucas-Tomasi (KLT).

Main Results:

• The proposed AMBP feature-based tracking method demonstrated superior performance compared to other descriptors.
• Achieved a high accuracy of 95% in real-world ultrasound data.
• The method proved effective in various challenging situations, including noisy and filtered images.

Conclusions:

• Presents the first fully automatic nerve tracking method specifically for ultrasound images.
• AMBP features offer superior performance for nerve tracking in ultrasound, outperforming other descriptors.
• The developed technique enhances the practice of Ultrasound-Guided Regional Anesthesia by improving nerve visualization and tracking.