Research into the Design and Validation of an Innovative Intraoperative Load Sensor for Total Knee Replacements through Experimental and Cadaveric Investigations.

Abstract

Balancing the force between compartments in the knee joint during a Total Knee Replacement (TKR) is a vital parameter for surgeons to achieve successful outcomes for patients due to the complexity of the joint, articulating surfaces, tendons, and ligaments. These structures are all impacted by the forces created in the joint following a TKR. A lack in equilibrium can cause an asymmetric gait, pain, a decrease in range of motion (ROM), loosening of the implant, and premature wear resulting in early revision surgeries. Load balancing involves the use of implants and tools to ensure the forces on the joint are evenly distributed across the articulating surfaces of the medial and lateral compartments. Surgeons typically rely on haptic feedback and experience to determine if the joint is balanced; however, the use of an objective device would help to quantify this objectively and ensure positive postoperative outcomes as opposed to traditional subjective methods. Current literature implies that such systems, which measure the load in the knee intraoperatively, require improvements to the accuracy and design before the impact can be further investigated. The design and validation of an objective tool to be used intraoperatively to balance the load in the knee was the subject of this research. The novelty of this system is in the gap in accuracy and sensitivity amongst existing systems on the market combined with design features which integrate well with the surgical workflow. Additionally, the compatibility of prior sensors with a variety of implant systems is a limitation of their use. Through this research the design of two sensors were fabricated and investigated. The design of these sensors included features to increase the sensitivity and accuracy for determining the magnitude of the load and its location. Finite Element Analysis (FEA) aided in design optimisation including strain gauge placement and material selection. Additionally, the development and use of Artificial Intelligence (AI) in this system was paramount. This was due to the complex geometry of the sensors meaning there was no closed form solution to determine the load and location of the femur’s articulation with the sensor in each compartment. Based on this nature of this problem an Artificial Neural Network (ANN) was used and was optimised based on physically collected training data. This research also aided in increasing the overall confidence of AI in the medical field which will aid regulatory bodies in their decision to allow for the clinical use of such devices to improve patient outcomes. A mixed method approach was used that included both quantitative and qualitative assessments of the design and the in-service use of the sensors. This included laboratory testing to investigate the performance of the AI in predicting the load and location of the contact force. Accuracy testing uncovered an average accuracy of about 90% for the Zimmer Specific sensor and about 88% for the Ring sensor when predicting the load. When predicting the centre of pressure, the average distance of the predictions from the actual location was 5.30 mm and 4.39 mm for the Zimmer Specific and Ring sensors respectively. Moreover, cadaveric testing improved the design of the sensor from the perspective of an experienced orthopaedic surgeon as well as confirmed the usability and proper function by comparing results to expected kinematic and kinetic behaviours of the knee.