Accuracy and Reliability of a Novel IMU-Based Functional Calibration Algorithm for Clinical 3D Wrist Joint Angle Monitoring

Inertial measurement units (IMU) represent a rapid advancement in biomechanical motion analysis. Still, their internal axis reference frame needs to be aligned with the anatomy to provide clinically meaningful metrics. The literature presents several sensor-to-segment calibration algorithms that can serve this purpose. However, due to the lack of wrist-specific calibration algorithms that can be universally applicable in rehabilitation, we aim to present and validate a functional calibration algorithm to monitor 3D wrist joint angles that can fill this gap. We recruited 13 healthy subjects with no sign of musculoskeletal injury and instructed them to perform wrist calibration movements, as well as a sequence of single-plane and multi-plane tasks. We conducted a statistical analysis using statistical parametric mapping (SPM) on continuous joint angle curves and traditional scalar statistical analysis on root mean squared error (RMSE) and range of motion (ROM) to compare the performance of our IMU calibration with an optical motion capture system (OPTO). SPM one-way MANOVA across different movements detected significant differences between OPTO and IMU between 0-20% (p=0.007) and 80-100% (p=0.002) of the movement; however, post-hoc ANOVA with Bonferroni correction did not reach significance (p>0.005). Scalar statistical analysis using one-way MANOVA on RMSE did not detect significant differences between the two systems (p=0.247). The maximum RMSE detected was 15.1° ± 9.4° during a drinking task on the radial/ulnar deviation axis, whereas the minimum was 3.9° ± 5.3° on the internal/external rotation angle during a drawing task. We conclude that our IMU-based FC system can be used as a basis for a wrist joint motion analysis or rehabilitation system that showed no significant differences with respect to a gold standard optical motion analysis system.