QUANTIFYING JOINT GEOMETRY IN HUMAN HANDS FROM IMAGING DATA

Abstract

INTRODUCTION

The geometry of the same joint varies substantially between people. Typical variation in merely how conforming the two subchondral bone surfaces are can increase the peak compressive stress on the articular cartilage by as much as the additional loading from becoming obese will. The mechanical environment of joint tissues is considered to play a central role in OA development. Quantifying joint geometry using repeatable, reliable, and accessible metrics supports better understanding of the relative importance (or unimportance) of this source of variability between people on their individual OA risk and this factor’s role at a population level.

OBJECTIVE

Previous methods of quantifying joint congruence (a measure of how conforming two surfaces are) have required detailed mathematical descriptions of the articulating surfaces and their relative position. We have developed a new method of measuring joint congruence that works directly from the 3D segmented point clouds. This has been applied to a joint in the thumb.

METHODS

The first step of the new methodology involves performing a Finite Element (FE) simulation of an elastic layer compressed between each set of segmented bones (Figure 1). The results of this are then interpreted using the elastic foundation model (Figure 2), enabling an equivalent, but far simpler, contact geometry to be identified. This far simpler equivalent geometry takes the form of a sphere contacting a flat surface. The identified congruence metric is the radius of this sphere, the ‘equivalent radius’, which produces an equivalent contact to that identified in each FE simulation. The minimal JSW (in this joint position) can also be estimated from the FE simulations. The new method has been applied to a small sample (n = 10) of healthy instances (5M:5F, mean age 31yrs) of the thumb metacarpophalangeal (MCP) joint (IRAS Ethics Ref: 14/LO/1059). Each participant’s right hand was CT scanned with near-isotropic voxel size (0.293 × 0.293 × 0.312 mm) and the bones segmented using a greyscale threshold.

RESULTS

To enable an appropriate reduction of the complex geometry represented in the 3D points clouds to one number (the radius of an equivalent ‘ball on flat’), this single parameter must continue to capture the joint’s geometry as the contact area increases. For all thumb MCP geometries tested, the force-displacement response of the elastic layer could be well-described by an identified equivalent radius, unique to that particular joint (Figure 3). The thumb MCPs had a mean equivalent radius of 17.9 mm (SD = 10.6 mm) and mean minimal JSW of 0.86 mm (SD = 0.24 mm). No relationship between congruence and joint space width was observed (Figure 4).

CONCLUSION

The new method can perform an efficient quantification of congruence, reducing two 3D point clouds to a single parameter. However, further application of the method has been postponed until questions around the role of CT/MRI scan resolution and the spatially varying geometry of articular cartilage have been explored in more detail. Initial results examining these questions using a μCT dataset of hands can be shared (Figures 5 and 6).