Computer Vision and Augmented Reality Guided Shoulder Rehabilitation.

Area of Science:

• Biomedical Engineering
• Computer Vision
• Rehabilitation Medicine

Background:

• Musculoskeletal disorders (MSDs) of the shoulder require accurate range of motion (ROM) assessment.
• Traditional goniometric methods can be cumbersome and less accessible for remote monitoring.
• Need for cost-effective, noninvasive tools for evaluating shoulder rehabilitation progress.

Purpose of the Study:

• To develop and validate a computer vision (CV)-based system for measuring shoulder ROM in rehabilitation.
• To provide a cost-effective and accessible alternative to traditional ROM measurement tools.
• To enhance patient engagement through real-time visual feedback during shoulder rehabilitation exercises.

Main Methods:

• Utilized the MediaPipe Pose model to predict shoulder joint angles.
• Developed a CV system for remote rehabilitation assessment of shoulder ROM.
• Implemented real-time color-coded visual feedback based on achieved ROM angles (<80°, 80°-100°, >100°).

Main Results:

• The CV system demonstrated high agreement with standard ROM measurement systems (LoA within ± 2.5°, RMSE < 1).
• Real-time visual feedback mechanisms were integrated to motivate patients.
• The system's performance was validated against commercial motion tracking sensors for clinical applicability.

Conclusions:

• The CV-based system offers a novel, accurate, and accessible method for quantifying shoulder ROM in rehabilitation.
• Real-time feedback enhances patient engagement and potentially improves rehabilitation outcomes.
• This technology has significant implications for musculoskeletal disorder care and remote physiotherapy.