Research and Development, Aesculap AG, 78532 Tuttlingen, Germany.
Department of Orthopaedic and Trauma Surgery, Musculoskeletal University Center Munich (MUM), Campus Grosshadern, Ludwig Maximilians University Munich, 81377 Munich, Germany.
Sensors (Basel). 2022 Dec 29;23(1):348. doi: 10.3390/s23010348.
The success of kinematic analysis that relies on inertial measurement units (IMUs) heavily depends on the performance of the underlying algorithms. Quantifying the level of uncertainty associated with the models and approximations implemented within these algorithms, without the complication of soft-tissue artefact, is therefore critical. To this end, this study aimed to assess the rotational errors associated with controlled movements. Here, data of six total knee arthroplasty patients from a previously published fluoroscopy study were used to simulate realistic kinematics of daily activities using IMUs mounted to a six-degrees-of-freedom joint simulator. A model-based method involving extended Kalman filtering to derive rotational kinematics from inertial measurements was tested and compared against the ground truth simulator values. The algorithm demonstrated excellent accuracy (root-mean-square error ≤0.9°, maximum absolute error ≤3.2°) in estimating three-dimensional rotational knee kinematics during level walking. Although maximum absolute errors linked to stair descent and sit-to-stand-to-sit rose to 5.2° and 10.8°, respectively, root-mean-square errors peaked at 1.9° and 7.5°. This study hereby describes an accurate framework for evaluating the suitability of the underlying kinematic models and assumptions of an IMU-based motion analysis system, facilitating the future validation of analogous tools.
基于惯性测量单元（IMU）的运动学分析的成功在很大程度上依赖于底层算法的性能。因此，量化这些算法中所实现的模型和近似方法相关的不确定性水平非常关键，而无需考虑软组织伪影的复杂性。为此，本研究旨在评估与受控运动相关的旋转误差。在这里，使用先前发表的荧光透视研究中的六位全膝关节置换患者的数据，通过将 IMU 安装到六自由度关节模拟器上来模拟日常活动的真实运动学。测试并比较了一种基于模型的方法，该方法涉及扩展卡尔曼滤波，以从惯性测量中推导出旋转运动学，与地面真实模拟器值相对比。该算法在估计水平行走过程中的三维旋转膝关节运动学方面表现出出色的准确性（均方根误差≤0.9°，最大绝对误差≤3.2°）。尽管与下楼梯和从坐到站再到坐相关的最大绝对误差分别上升至 5.2°和 10.8°，但均方根误差的峰值分别为 1.9°和 7.5°。本研究描述了一种评估基于 IMU 的运动分析系统中底层运动学模型和假设适用性的精确框架，为类似工具的未来验证提供了便利。
