Estimation, Search, and Planning (ESP) Research Group

Background Isolated lateral compartment osteoarthritis (OA) occurs in ~10% of arthritic knees. These patients are candidates for a unicompartmental knee replacement with an Oxford Domed Lateral implant (ODLUKR). However, 1 to 6 % of patients who receive this implant will experience dislocation of the mobile bearing. Clinical observations suggest that although dislocations can occur anteriorly or posteriorly, medial dislocations, where the bearing becomes lodged on the wall of the tibial component, are by far the most common type. The mechanism behind mobile bearing dislocation is not understood but anecdotally they are said to occur when the knee is flexed and unloaded. We have explored the use of the Open Motion Planning Library (OMPL) platform, an open-source path planning library commonly used in robotics, to assess the circumstances under which dislocation of the mobile bearing in the Oxford Domed Lateral implant can occur.

Objectives The aim of this study was to understand the mechanism of mobile bearing dislocation by developing a tool using methods commonly employed in robotics. We would like to use this tool to inform the design of new implants, which will reduce the incidence of mobile bearing dislocation.

Study Design & Methods OMPL requires the definition of an ‘environment’ through which the ‘robot’ can navigate. The Oxford Domed Lateral implant femoral and tibial components were used to define the environment. The environment was altered parametrically by moving the components relative to one another i.e. vertical (0–8 mm) and mediolateral (0–4 mm) translation distances in 0.25 mm increments. To analyse the ability of the mobile bearing to move within this environment from its start to a region defining dislocation (“goal area”). Dislocation was achieved when the centre of the bearing fell within this defined goal area. To analyse the ability of the bearing to reach the goal area for each configuration, the RRT path planning algorithm was applied to the bearing for up to 60 seconds and 10 iterations. The problem was confined by a bounding box to bias the solution in the medial direction, to align with clinical findings. The results from the model were compared to measurements taken manually on a mechanical rig.

Results Preliminary testing of the platform has showed that application of robotics path planning algorithms to this problem is feasible. Compared with dislocation results obtained using a previously built mechanical rig, there is good correlation (r² = 0.83) between the results obtained manually and using this robotics analysis tool. Results from the robotics analysis tool show that medial dislocation occurs at 4.75 mm of vertical distraction when mediolateral translation is 0 mm. This value reduced to 3.75 mm when there is 4 mm of mediolateral translation.

Conclusions The robotics analysis tool demonstrates a successful application of the robotics path planning algorithm to the clinically relevant problem of mobile bearing dislocation in the Oxford Domed Lateral implant. Results from the robotics analysis tool show a surprisingly small amount of vertical distraction required for medial dislocation. This distraction amount is far smaller than the amount that the lateral compartment of the knee joint can distract, particularly in flexion, when there is increased ligament laxity; and may explain why dislocations are more frequent in the lateral knee. This model will also be used as a testing tool to identify the appropriate new implant design, which is less prone to dislocation. If successful, the new implant can reduce revision surgery rates and improve patient outcomes.