Accuracy of the Automated Range of Motion Observer and Reporter Software for Fully Automated Joint Measurement From Patient Videos.

Abstract

Accurate assessment of joint range of motion (ROM) is essential for diagnosing and managing upper extremity injuries. Universal goniometers are the most used tools for measuring ROM, but they require skilled observers and are limited by interobserver variability. An automated system for measuring joint range of motion from patient videos could facilitate standardized reporting of outcomes after reconstructive surgery. An Automated Range of Motion Observer and Reporter (ARMOR) software was developed as an autonomous, video-based ROM measurement tool leveraging OpenCV pose estimation. ARMOR was used to assess upper extremity range of motion and was validated against photography-based (photogoniometry) and manual goniometry in a cohort of brachial plexus surgery patients. The correlation coefficients comparing ARMOR to manual goniometry were above 0.90 for all motion tasks, except for elbow flexion. For shoulder flexion, the mean difference between ARMOR and manual goniometry was more than 14° smaller than the difference for photogoniometry. Mean differences for shoulder abduction were similar between ARMOR and photogoniometry. ARMOR's accuracy in assessing shoulder ROM, independence from human observer bias, and telemedicine compatibility makes it a promising solution for consistent and accessible ROM assessment. The autonomous nature of the software enhances the data collection workflow for clinical researchers while eliminating interrater variability.