Jump to Content
  1. Oxford
  2. MPLS
  3. Eng. Sci.
  4. ORI

Estimation, Search, and Planning (ESP) Research Group

A novel computational dislocation analysis tool using robotics path planning algorithms for mobile bearing lateral unicompartmental knee replacement surgery

Authors
  1. Irene Yang
  2. Jonathan D. Gammell
  3. David W. Murray
  4. Stephen J. Mellon
Publication Date
Abstract

Introduction Ligament laxity in the lateral knee enables dislocation of the mobile bearing in 1–6% of Oxford Domed Lateral (ODL) Unicompartmental Knee Replacements (UKRs). Although dislocations can occur anteriorly or posteriorly, they usually occur medially with the bearing sitting above the wall of the tibial component. Dislocations were previously studied using a custom built mechanical rig, however, testing using the rig was inefficient. The aim of this study was to develop a more reliable and efficient dislocation analysis tool.

Methods The established robotics software package, the Open Motion Planning Library (OMPL) was modified to accept the ODL components. Starting with the components positioned with the bearing flush against the tibial wall, the femoral component was distracted away from the tibial component: vertically (2–8 mm) and mediolaterally (ML) (0–6 mm) in 0.25 mm increments (425 configurations). Using a robotics path planning algorithm called Rapidly-Exploring Random Trees (RRT) the mobile bearing was allowed to find a way out from between the femoral and tibial components i.e. to dislocate. Dislocations were labelled as medial, lateral, anterior or posterior depending on the dislocation direction. For each ML distance, the vertical distraction required for dislocation (VDD) was recorded. To improve the tool’s efficiency, a convergence test was run to identify the optimal search time, no. of searches. The tool was also made more agile by implementing directional searching and selectively increasing the search capacity. To validate the tool, the VDD results were compared to measurements taken using a custom-built mechanical rig. Medial VDD results were statistically compared using a Intraclass Correlation Coefficient: mean rating (k=2), two way random effects model with type consistency and 95% Confidence Interval (CI).

Results Section Using the RRT algorithm, the tool successfully identified dislocations medially, laterally, anteriorly and posteriorly. Convergence testing found that at least 10 search attempts and 210 seconds was required to identify all solutions. However, within 10 search attempts and 45 seconds search time, most of the dislocation solutions had been found. Based on this, for each configuration, the RRT algorithm was allowed to run for 270 seconds and 10 searches. If no solution was found in this time, the search was repeated for the same configuration but with the time and search attempts extended to 405 seconds and 25 searches, respectively. For medial VDD, as ML translation increased from 0 mm to 6 mm, the VDD medially of the mobile bearing reduces: 5.5 mm to 2.75 mm for the mechanical rig and from 5.5 mm to 3.25 mm for the robotics tool. For lateral, anterior and posterior dislocations, the robotics dislocation analysis tool output VDDs which were marginally higher than the manual mechanical rig: 3.5 mm, 6.25 mm and 6.25 mm versus 2.75 mm, 5.75 mm and 6 mm respectively. The ICC for medial VDD was 0.993 (95%CI: 0.982–0.998).

Discussion Our study demontrates a novel and successful application of a robotics path planning algorithm, RRT, to the clinical problem of mobile bearing dislocation, to develop a reliable and efficient robotics dislocation analysis tool. The results from the tool reveal that the amount of distraction for a medial dislocation (0–6 mm ML: 5.5–3.25 mm) was much smaller than that for an anterior or posterior dislocation (6.25 mm). This explains why medial dislocations are more common clinically. Even though the amount of distraction required for lateral dislocation was even less than that required for a medial dislocation, when the lateral compartment is distracted the lateral ligament and other soft tissues are tight, preventing a lateral dislocation. Anterior and posterior bearing dislocation rarely occur and this dislocation rate is clinically acceptable. Future work will use the tool to test whether modifications to the design of the implant e.g. increasing the height of the tibial component wall successfully reduce the risk of medial dislocations. More broadly, the tool may be used to inform the implant design to reduce medial dislocation risk to match that of anterior/posterior dislocation, which would reduce the dislocation risk to an acceptable level.

Significance/Clinical Relevance Using the novel dislocation analysis tool developed in this study, medial dislocation required less distraction than either anterior or posterior dislocation, possibly explaining why medial dislocations are most common clinically. If the VDD medially could be reduced to match the VDD anteriorly or posteriorly, the dislocation risk would likely be acceptable.

Publication Details
Type
Abstract-Refereed Conference Paper
Conference
Orthopedic Research Society (ORS) Annual Meeting
Location
Tampa, FL, USA
BibTeX Entry
@inproceedings{yang_ors22,
author = {Irene Yang and Jonathan D Gammell and David W Murray and Stephen J Mellon},
title = {A novel computational dislocation analysis tool using robotics path planning algorithms for mobile bearing lateral unicompartmental knee replacement surgery},
booktitle = {Proceedings of the Orthopedic Research Society ({ORS}) Annual Meeting},
year = {2022},
address = {Tampa, FL, USA},
month = {4--8 } # feb,
}