Doerks Frithjof, Harms Fenna, Schwarze Michael, Jakubowitz Eike, Welke Bastian
Hannover Medical School, Department of Orthopaedic Surgery, DIAKOVERE Annastift, Laboratory for Biomechanics and Biomaterials, Hannover, Germany.
Department for Medical Technology, Bremerhaven University of Applied Sciences, Bremerhaven, Germany.
PLoS One. 2025 May 30;20(5):e0324499. doi: 10.1371/journal.pone.0324499. eCollection 2025.
Although three-dimensional marker-based motion analysis is the gold standard for biomechanical investigations, it is time-consuming and cost-intensive. The conjunction of monocular video recordings with pose estimation algorithms addresses this gap. With the Orthelligent VISION app (OPED GmbH) a commercial and easy-to-use tool is now available for implementation in everyday clinical practice. The study investigates the accuracy of the 2D video-based system in measuring joint kinematics, expressed as range of motion, compared to an optoelectronic 3D motion analysis system as the gold standard.
Its accuracy was determined by synchronously measuring ten healthy subjects with Orthelligent and the optoelectronic 3D motion analysis system Qualisys (Qualisys AB) during level walking and at different treadmill walking speeds (1 m/s; 1.4 m/s; 1.8 m/s). Range of motion (RoM) of lower limb joints and time-distance parameters were compared using Bland-Altman plots, t-tests, and correlations between systems. Kinematic outputs of two subjects with a lower limb amputation were also analyzed.
The mean RoM deviation was smaller for the knee (3.8°) and hip joints (3.7°) than for the ankle joint (5.4°), but differed significantly between systems in most conditions. The correlation range was 0.36 ≤ r ≤ 0.83, with best results for 1 m/s treadmill walking (mean r = 0.71 across joints). While the accuracy was affected by high inter-subject variability, individual RoM changes from slow to fast walking did not differ between the systems. The kinematics of the prosthetic and sound leg of individuals with an amputation exhibited characteristic patterns in the video-based system, even though side differences were smaller compared to the optoelectronic measurement.
The rather high inter-subject variability would make future comparisons between individuals challenging. Nonetheless, the app shows potential for intra-subject progress monitoring.
尽管基于三维标记的运动分析是生物力学研究的金标准，但它既耗时又成本高昂。单目视频记录与姿势估计算法的结合弥补了这一差距。借助Orthelligent VISION应用程序（OPED GmbH），现在有了一种商业且易于使用的工具可用于日常临床实践。本研究调查了基于二维视频的系统在测量关节运动学（以运动范围表示）方面的准确性，并与作为金标准的光电三维运动分析系统进行比较。
通过在平地上行走以及不同跑步机行走速度（1米/秒；1.4米/秒；1.8米/秒）下，使用Orthelligent和光电三维运动分析系统Qualisys（Qualisys AB）同步测量十名健康受试者来确定其准确性。使用布兰德 - 奥特曼图、t检验以及系统之间的相关性来比较下肢关节的运动范围（RoM）和时间 - 距离参数。还分析了两名下肢截肢受试者的运动学输出。
膝关节（3.8°）和髋关节（3.7°）的平均RoM偏差小于踝关节（5.4°），但在大多数情况下系统之间存在显著差异。相关范围为0.36≤r≤0.83，在跑步机以1米/秒速度行走时效果最佳（各关节平均r = 0.71）。虽然准确性受个体间高变异性影响，但系统之间个体从慢走到快走时RoM的变化并无差异。截肢个体的假肢和健全腿的运动学在基于视频的系统中呈现出特征模式，尽管与光电测量相比侧别差异较小。
个体间相当高的变异性将使未来个体之间的比较具有挑战性。尽管如此，该应用程序在个体内进展监测方面显示出潜力。
