Computer Vision and Augmented Reality Guided Shoulder Rehabilitation

Range of motion (ROM) measurements are essential for evaluating and diagnosing musculoskeletal disorder (MSDs) in the shoulder, such as rotator cuff tendinitis, impingement, rotator cuff tears, and tendinopathy. This study proposes a computer vision (CV)-based system for shoulder rehabilitation ROM measurement as a cost-effective, noninvasive, and accessible alternative to traditional goniometric methods. The system uses the MediapPipe Pose model to predict shoulder joint angles for remote rehabilitation assessment. The predicted joint angles were in agreement (LoA within ± 2.5° and RMSE < 1) with standard ROM measurement systems. Real-Time visual feedback with dynamically drawn trajectories on the screen around the subject are color-coded as red, orange, and green to indicate achieved ROM angles of <80°, between 80° and 100°, and >100°, respectively. These feedback mechanisms enhance patient engagement by motivating them to improve their rehabilitation outcomes. Thus, a novel system has been designed exclusively for shoulder rehabilitation by quantifying their performance using ROM metrics. The results have been compared against commercially available motion tracking sensors towards validating its real-time clinical applicability.Clinical Relevance- The CV-based joint movement analysis has significant implications, particularly for MSD care and rehabilitation, by providing real-time feedback to physiotherapists and patients.