Abstract
Summary
Our statistical shape analysis showed that size is the primary geometrical variation factor in the medial meniscus. Shape variations are primarily focused in the posterior horn, suggesting that these variations could influence cartilage contact pressures.
Introduction
Variations in meniscal geometry are known to influence stresses and strains inside the meniscus and the articulating cartilage surfaces. This geometry-dependent functioning emphasizes that understanding the natural variation in meniscus geometry is essential for a correct selection of allograft menisci and even more crucial for the definition of different sizes for synthetic meniscal implants. Moreover, the design of such implants requires a description of 3D meniscus geometry. Therefore, the aim of this study was to quantify 3D meniscus geometry and to determine whether variation in medial meniscus geometry is size or shape driven.
Patients & Methods
Sagittal knee MR images (n=35; 15 males, 20 females, aged 33±12) were acquired at 3 Tesla using a 3D SPACE sequence with isotropic resolution of 0.5×0.5×0.5mm. 3D models were generated by manual segmentation of the medial menisci from the MR scans. The surface of a reference meniscus was then described by 250 landmarks. Using an affine iterative closest point transformation, these landmarks were registered onto the full set of 3D models. Based on the set of corresponding landmarks, a point distribution model was created using the Shapeworks software (NITRC, University of Utah), an open source algorithm for constructing correspondence-based statistical models of sets of similar shapes. Several modules from Shapeworks and the Arthron software (UCD, Dublin) were used to perform principal component analysis (PCA) upon the set of landmarks. The results of the PCA enabled quantification and visualisation of the primary modes of variation in meniscal geometry.
Results
The majority (77%) of variation in medial meniscus geometry was found to be due to sizing (principal component (PC) 1). Including the shape-related PC's 2 to 4, increased the cumulative percentage of represented geometry variation to over 90%. The independent shape variations described by PCs 2–4 all display larger variations in geometry of the posterior meniscal horn than the anterior section.
Discussion
From this study, we can conclude that geometry variation of the medial meniscus is mainly determined by differences in size. However, since the posterior aspect of the medial meniscus experiences higher loads during daily activities than the anterior part, the shape variations described by PCs 2–4 may have a significant influence on cartilage contact pressures. Therefore, PCA alone does not provide sufficient information to define the number of implant sizes to cover a majority of the population. Analysis of the sensitivity of cartilage contact pressures to the shape variations identified in this analysis could provide the additional information needed.
