Inertial-Based 3D Knee Joint Angular Kinematics Estimation: Effect of Number of Repetitions During Functional Calibration

: Inertial-based joint kinematics is gaining significant attention due to the compactness and low cost of inertial measurement unit (IMU). Anatomical reference frames from IMUs can be defined considering specific subject posture and movements; this procedure is known as functional approach. Typically, movements are repeated a certain number of times, mainly ten times, to improve the accuracy of the estimated frames. However, no studies have investigated whether fewer repetitions result in a significant degradation of the angle estimates. Thus, the main aim of the present work is to present an anatomical calibration approach based on subject movements to define the anatomical reference frames reducing gradually the duration of the considered trial used for the calibration. Based on the results, knee joint 3D kinematics has been estimated including one, three and ten repetitions of the calibration movements, and then compared with the gold-standard optoelectronic system. Errors in the estimates were assessed using the root mean square (RMSE) and the mean absolute percentage (MAPE) errors. The best results have been obtained including ten repetitions, RMSE of 2.1 deg, 3.8 deg, and 4.6 deg for flexion/extension (FE), abduction/adduction (AA) and internal/external (IE) rotation, respectively, outperforming other studies. Notably, these results were not statistically different from the condition with three repetitions for the FE, while a strong difference has been observed for the IE. These findings support the feasibility of obtaining clinically relevant kinematics from IMUs with reduced time for the functional calibration, thus improving patient comfort and minimizing fatigue.Clinical relevance- Presented outcomes allow the possibility to obtain reliable kinematics estimates with non-obtrusive sensors reducing the required time for sensor calibration, also improving patient comfort during evaluation session.