Optimizing MR Image Reconstruction of the Knee Joint for Computational Simulation

Computational models can be useful tools in helping to make clinically relevant decisions, such as assessing the likelihood of injury or optimizing surgical intervention for a patient. For adequate computational evaluation of patellar dislocation in a high risk population, detailed subject-specific 3-D models are required. However, clinical scans available for this patient population are typically very coarse (slice thickness~3mm). The purpose of this study is to efficiently create 3-D subject-specific models from course scans such that they can be used in computational simulations.

Commercially available software (Amira, FEI, OR) is used to generate 3-D models of the knee joint from clinical magnetic resonance images (MRI) from a group of patients suffering from recurrent patellar dislocation. The geometric reconstruction must be sufficiently detailed for use in finite element (FE) simulations (Abaqus, Simulia, RI). A set of clinical scans were reconstructed with varying levels of coarseness, ranging from 29 slices to 300 slices. Then each model version will be incorporated into an FE simulation of a squat activity and knee joint mechanic outputs (kinematic contact mechanics) will be compared across all sample models to determine the minimum amount of refinement required to produce consistent FE results.