The Application of Artificial Intelligence in Wrist and Hand Joint Ultrasound

Sichuan University

Status: Enrolling

Conditions: Education; Joint Diseases

Study type: Observational

Funder types: Other

Identifiers

NCT06883669

2021(1010)

Details and patient eligibility

About

1. To develop an AI system that can automatically identify standard sections and save images during wrist and hand joint ultrasound scans, while labeling key anatomical structures.
2. To recruit sonographers untrained in musculoskeletal ultrasound, train them in wrist and hand joint scans, and compare their scanning speed and image quality when using and not using the AI system.

Full description

Research Background Musculoskeletal ultrasound, a rapidly evolving technique for ultrasound - based diagnosis and treatment of the musculoskeletal system, has seen expanding applications in visualizing peripheral nerves, muscles, tendons, joints, and skin thanks to improved ultrasound resolution. It offers advantages like convenience, cost - effectiveness, safety, real - time dynamics, continuous follow - up, and fast reporting. Given the high incidence and wide prevalence of rheumatic and immunological diseases, high - frequency ultrasound is gaining clinical attention. Consequently, learning and promoting musculoskeletal ultrasound is clinically valuable and necessary. However, this field faces challenges such as operator dependence, slow skill improvement and subjective differences in ultrasound diagnosis criteria and assessment methods for rheumatic diseases. The emergence of intelligent tools (AI) can address the urgent need for fast, accurate, and standardized ultrasound diagnosis. While AI has been used in multiple ultrasound sub - specialties, its application in musculoskeletal ultrasound is limited. This study aims to develop an AI - assisted musculoskeletal ultrasound examination system. It will help ultrasonographers by real - time segmenting structures like muscles, tendons, bone cortex, peripheral nerves, joint spaces, and blood vessels, extracting lesions (e.g., synovitis, tenosynovitis, bone erosion, cartilage destruction), and assessing lesion severity, thereby improving diagnostic accuracy and efficiency. Additionally, the system will shorten the learning cycle, enhance learning efficiency for musculoskeletal ultrasound, and accelerate its adoption in hospitals at all levels.

Research Objectives Primary Objective: To develop a musculoskeletal ultrasound AI - assisted examination system. This system will identify standard examination sections and continuous dynamic images of wrist and hand joints, mark specific structures in real - time, extract lesions, and assess their severity, enhancing examination efficiency and accuracy.

Secondary Objective: To help ultrasonographers shorten their learning period and master musculoskeletal ultrasound examination skills more quickly through the application of the AI - assisted system.

Enrollment: 500 estimated patients

Sex: All

Ages: 18 to 100 years old

Volunteers: Accepts Healthy Volunteers

Inclusion and exclusion criteria

(1) Research Subjects-for AI system establishment

Inclusion Criteria:

1. Healthy volunteers with good compliance
2. No history of disease on peripheral nerve, muscle, or tendons;
3. No early - stage RA or history of RA.

Exclusion Criteria:

1. Amputees or those with limb disabilities.

Individuals with poor compliance.

(2) Research Subjects-for AI system validation

Inclusion Criteria:

1. Sonographers with at least 2 years of ultrasound scanning experience

Exclusion Criteria:

1. Those who have performed musculoskeletal ultrasound scanning or received related training

Trial design

500 participants in 2 patient groups

Research Subjects-for AI system establishment

Description:

Collect ultrasound images of wrist and hand joints, annotate and segment them frame - by - frame for model training.

Research Subjects-for AI system validation

Description:

Recruit sonographers untrained in musculoskeletal ultrasound, train them in wrist and hand joint scans, and compare their scanning speed and image quality when using and not using the AI system.