Clinical Evaluation of a Markerless Motion Capture Technology: Reliability and Usability for Musculoskeletal Assessment

BACKGROUND While various technologies have recently emerged as prominent tools in healthcare, particularly for augmenting screening and diagnostic processes, evaluations of their accuracy and reliability within clinical musculoskeletal (MSK) settings remain limited. METHODS This was a cross-sectional clinical evaluation and usability study of a markerless motion capture system (MMCS) compared with the standard of care using a manual goniometer in a healthy adult population. Ten range of motion (ROM) measurements (degrees) were taken on the same participant using both methods. Active, passive, and 3-repeat active joint ROM measurements were taken for the right and left sides. Mean paired differences between left and right pairs of MMCS and standard of care measurements were calculated and statistically compared using a paired t-test. Quintiles for each pair of ROM measurements were determined. Agreement of pair of ROM measurements was assessed using the cross-tabulations of the quintiles. Cohen’s kappa statistic, a measure of inter-rater reliability (comparing categories of manual goniometer vs. MMCS) was calculated. Intra-class correlation coefficient (ICC) was used to assess the reliability of each joint pair of ROMs. Mean absolute differences in ROM between joint pairs were calculated. A two-sided p-value RESULTS For active, passive, and 3-repeat active range of motion (ROM) showed fair agreement between manual goniometer and MMC measurements for active ROM, with kappa statistics ranging from 0.32 to 0.37. The ICCs varied across joints, ranging from .121 (active ROM left thoracolumbar rotation) to .960 (3-Repeat active ROM left shoulder internal rotation). Paired t-test of ROMs revealed statistically significant differences between the two methods for all joints except for active knee extension. About 40% of the mean absolute differences between ROM pairs were within ±5 degrees. CONCLUSION Mean differences between pairs of ROM measurements for manual goniometer vs. MMCS varied widely. Inter-rater reliability showed poor to excellent agreement. Reliability assessed using the ICC ranged from weak to strong. Future studies evaluating MMCS for screening purposes rather than diagnostic-level accuracy are needed. The reliability of the MMCS algorithm needs improvement before the technology can be scaled up.