The Developing of a Smart Elbow Prosthesis for Loosening Detection

Abstract

Total Elbow Arthroplasty (TEA) is an effective surgical procedure for restoring elbow joint function and improve a patient's quality of life by relieving pain suffered from various musculoskeletal disorders. Despite new designs for prostheses and improved surgical procedures, TEA still suffers today from mid to-long-term complications such as aseptic loosening, infection, dislocation, and pre-prosthetic fractures. With aseptic loosening followed by infection being the most persistent reason for TEA revision, investigating methods for early diagnosis of implant loosening and differentiating between the infection and aseptic loosening is necessary to address this problem. This thesis aims to develop a novel diagnostic tool that can be embedded into the prosthetic and provide a quantitative measurement for early signs of the implant loosening without any usage of radiographs or any contact with the implant. In this study, three types of sensor configurations along with detection algorithms were developed, designed, and tested along with a functional prototype to detect the migration of the elbow prosthesis (Aseptic loosening). The detection system was validated under realistic conditions through experiments with a custom-designed mechanical testing rig. Finally, for infection detection, a biocompatible chemical sensor (Hydrogel) was synthesised and was linked with the aseptic loosening detection system to investigate the early signs of infection. Among the three sensor configurations, the single sensor configuration detected the implant migration at a resolution of 0.3 mm with a detection error of less than 3 %. The configuration was able to detect angular motion up to 3 degrees with a detection error of 5 %. The quad sensor configuration, an arrangement of four closely packed sensors, enhanced the overall detection performance by increasing system resolution to 0.15 mm in multiple axes along with increasing the signal to noise ratio, reducing root mean square error, and compensating the tilt effect of the single sensor. While the dual sensor configuration, two sensors arranged in-line but 42 mm apart, downgraded the detection performance by introducing a detection error of 30 %. The detection system showed negligible effect on the biomaterial used in TEA and was able to differentiate between different migrations types (Linear, Angular, Static and Dynamic). The difference in three fixation scenarios (grossly loose, partially loose, and fully fixed) was identified evidently by the detection system with the grossly loose fixation showed a displacement of 0.187 ± 0.061 mm on the x-axis and 0.387 ± 0.059 mm on the y-axis. The chemical sensor (Hydrogel) was able to detect the change in its surrounding pH level (highlighting the potential to detect infection) and by the amalgamation with the detection system, pH change was detected without the use of an imaging technique. Further improvement in the synthesis of the hydrogel and the optimisation of the detection system has also been suggested. The quad sensor system implies that it has the potential to be used to continually or intermittently monitor implant behaviour without hospital visitation or x-ray exposure. This could be applied more widely to other major joints such as the hips and knees, giving in-situ biomechanical insight into joint replacement behaviour over time.