Feasibility study of a sensor-to-segment calibration method to enhance upper limb motion analysis using an IMU-based system for clinical and home environments.

Inertial Measurement Units (IMUs) represent a valid alternative to standard clinical assessment methods, such as clinical scales, for evaluating upper limb kinematics. A key aspect of utilizing IMUs effectively is ensuring precise sensor-to-segment calibration, which accounts for the relative orientation between the sensor and the attached body segment. This calibration is crucial to obtain accurate results. Although reliable calibration methods are available, their application in clinical and home environments remains challenging due to their complexity. This study aimed to validate a picture-based calibration method feasible for a clinical context and compare it against other standard methods. Ten healthy subjects performed daily activity tasks while upper limb kinematics was recorded using an optoelectronic motion capture system and an IMU-based system. Four calibration methods were compared using error metrics, including root mean square deviation (RMSD) and cross-correlation (XCORR). The results demonstrate that the proposed picture-based method provides highly accurate measurements for the first and second Euler rotation angles of the shoulder, with RMSD < 15° and XCORR > 0.75 across most of the tasks. For the elbow joint, all calibration methods consistently yielded precise results for the first rotation (RMSD < 15° and XCORR > 0.95) across the majority of tasks. The proposed sensor-to-segment calibration method improves the accuracy of upper limb motion data recorded with an IMU-based system compared to traditional methods. Moreover, the calibration approach is easy to use, making it suitable for clinical and home environments.