Related Experiment Video
Updated: Jun 10, 2026
06:44
Development of a Novel Task-oriented Rehabilitation Program using a Bimanual Exoskeleton Robotic Hand
Published on: May 20, 2020
AI-enhanced Wrist-Hand US Image Acquisition: Development and Initial Clinical Evaluation
Xinyi Tang1, Yujia Yang1, Lingyan Zhang1
1Department of Medical Ultrasound, West China Hospital, Sichuan University, No. 37 Guo Xue Xiang, Chengdu 610041, China.
Radiology
|June 9, 2026
Summary
This study developed artificial intelligence (AI) models for musculoskeletal ultrasound (MSK US) to improve image analysis. The AI tool enhanced standard plane recognition and anatomical segmentation, boosting novice sonologists' accuracy and efficiency in wrist-hand examinations.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Musculoskeletal Ultrasound
Background:
- Musculoskeletal (MSK) ultrasound examinations are crucial but highly subjective, relying heavily on sonologist expertise.
- Developing objective tools for MSK US can improve diagnostic consistency and accessibility.
Purpose of the Study:
- To create and evaluate deep learning models for standard plane recognition and anatomical segmentation in dynamic wrist-hand ultrasound examinations.
- To assess the clinical performance of these AI models in assisting novice sonologists.
Main Methods:
- A deep learning tool using Residual Network and High-Resolution Network was developed.
- The tool was trained on 66,743 normal wrist-hand US images from 430 volunteers.
- Clinical evaluation involved novice sonologists using AI-assisted versus conventional scanning, with efficiency and accuracy metrics analyzed.
Main Results:
- The AI tool achieved 96.2% F1 score for plane recognition, outperforming baseline models.
- It reached a median mean intersection over union of 0.647 for anatomical segmentation, surpassing other models.
- AI assistance significantly increased standard plane acquisition rates and reduced nonstandard images for novice sonologists, improving scanning efficiency.
Conclusions:
- The developed deep learning model demonstrates robust performance in standard plane recognition and anatomical segmentation for MSK ultrasound.
- AI assistance significantly improved the quality and efficiency of novice MSK sonologists during dynamic wrist-hand examinations.
