Deep visual nerve tracking in ultrasound images

Area of Science:

• Medical imaging
• Artificial intelligence
• Anesthesiology

Background:

• Ultrasound-guided regional anesthesia (UGRA) is standard for surgery and pain management.
• Accurate nerve localization is vital for UGRA, but challenging due to image noise and anatomical variability.
• Developing effective nerve tracking tools is essential for UGRA practice.

Purpose of the Study:

• To explore and comparatively evaluate recent deep learning (DL) trackers for nerve tracking in ultrasound images.
• To assess DL tracker performance in accuracy, consistency, and computational time.
• To analyze the ability of DL trackers to handle challenging nerve tracking scenarios like disappearance and shape distortion.

Main Methods:

• Thirteen recent deep learning trackers were investigated for nerve tracking applications.
• Trackers were evaluated on diverse ultrasound nerve images.
• Performance metrics included accuracy, consistency, time complexity, and robustness to image artifacts and nerve variations.

Main Results:

• Deep learning trackers demonstrated good overall performance in nerve tracking.
• Comparative analysis revealed varying strengths among the thirteen DL trackers.
• Certain DL trackers showed particular efficacy in handling challenging nerve tracking situations.

Conclusions:

• Deep learning trackers offer a promising solution for nerve tracking in ultrasound imaging.
• The study provides insights into the comparative performance of DL trackers for UGRA applications.
• Further development and application of DL trackers can enhance the precision and efficiency of UGRA procedures.