RESUMO
Understanding the loads that occur across musculoskeletal joints is critical to advancing our understanding of joint function and pathology, implant design and testing, as well as model verification. Substantial work in these areas has occurred in the hip and knee but has not yet been undertaken in smaller joints, such as those in the wrist. The thumb carpometacarpal (CMC) joint is a uniquely human articulation that is also a common site of osteoarthritis with unknown etiology. We present two potential designs for an instrumented trapezium implant and compare approaches to load calibration. Two instrumented trapezia designs were prototyped using strain gauge technology: Tube and Diaphragm. The Tube design is a well-established structure for sensing loads while the Diaphragm is novel. Each design was affixed to a 6-DOF load cell that was used as the reference. Loads were applied manually, and two calibration methods, supervised neural network (DEEP) and matrix algebra (MAT), were implemented. Bland-Altman 95% confidence interval for the limits of agreement (95% CI LOA) was used to assess accuracy. Overall, the DEEP calibration decreased 95% CI LOA compared with the MAT approach for both designs. The Diaphragm design outperformed the Tube design in measuring the primary load vector (joint compression). Importantly, the Diaphragm design permits the hermetic encapsulation of all electronics, which is not possible with the Tube design, given the small size of the trapezium. Substantial work remains before this device can be approved for implantation, but this work lays the foundation for further device development that will be required.