Related Experiment Video
Updated: Mar 30, 2026
Measuring 3D In-vivo Shoulder Kinematics using Biplanar Videoradiography
Published on: March 12, 2021
Real-time Automatic Guidance During Shoulder Ultrasound Scanning with Artificial Intelligence: Reducing Operator
Yanni He1, Zhenzhou Li2, Haorong Wu3
1Department of Ultrasound, Institute of Ultrasound in Musculoskeletal Sports Medicine, The Affiliated Guangdong Second Provincial General Hospital of Jinan University, Guangzhou, PR China (Y.H., H.W., W.Y., S.L., M.Z., T.B., J.Y., T.W., P.W., H.L.); The Second School of Clinical Medicine, Southern Medical University, Guangzhou, PR China (Y.H., H.L.).
An artificial intelligence (AI) system standardizes shoulder ultrasound (US) acquisition by providing real-time guidance. This AI tool reduces novice operator dependency and improves scanning efficiency, advancing musculoskeletal (MSK) US imaging.
Area of Science:
- Musculoskeletal imaging
- Artificial intelligence in medicine
- Ultrasound technology
Background:
- Shoulder ultrasound (US) scanning requires expertise, leading to variability in image acquisition.
- Standardizing the US scanning process is crucial for reliable diagnostic interpretation.
- Novice operators often struggle with consistent and accurate shoulder US examinations.
Purpose of the Study:
- To develop and validate an artificial intelligence (AI)-guided system for real-time automatic classification and structural recognition of shoulder US planes.
- To standardize the shoulder US scanning process through AI-driven guidance.
- To assess the clinical utility and performance of the AI system in real-world settings.
Main Methods:
- A prospective multicenter study utilizing convolutional neural networks for AI model development.
- Training and internal testing on 13,312 shoulder US videos (852 standard plane images, 74,909 frame images).
- External validation using 480 videos (8458 frame images) from a separate center, evaluating classification (AUC) and structure detection (mAP) metrics.
Main Results:
- The EfficientNetB2-based AI system achieved high performance on external validation (AUC: 0.99; mAP: 0.89), guiding 15 standard planes and localizing 27 structures.
- AI guidance significantly reduced shoulder US examination time for junior residents by 34% (10.06 min vs. 15.26 min, p=0.014).
- Real-time guidance accuracy was confirmed by expert evaluation, comparable to expert supervision.
Conclusions:
- The developed AI system effectively standardizes shoulder US acquisition through accurate, real-time guidance.
- The AI tool reduces operator dependency for novice sonographers, enhancing scan reproducibility.
- This represents a significant advancement towards standardized musculoskeletal (MSK) ultrasound imaging acquisition and interpretation.
