Optical measurement of the structure of articular cartilage

Medical conditions such as osteoarthritis are painful, debilitating, and hard to treat because they involve degradation of tissues that have a poor natural repair capacity, such as articular cartilage (which coats the sliding surfaces of articulating joints such as the knee). Cartilage is a material with remarkable mechanical properties that has evolved to tolerate the large shear and compressive stresses that are generated during everyday activities such as standing up and walking. Its mechanical properties arise from the interplay between the 3D collagen architecture, extracellular water, and proteoglycans (protein-polysaccharide conjugates that bind water). Arthritis involves a gradual destruction of the cartilage collagen architecture. Research into the properties of natural cartilage is vital to improving our understanding of the onset, progression, and treatment of arthritis and also in developing new treatment strategies, for example, through tissue engineering.

Figure 1. A 3D model of the collagen fiber orientation in healthy cartilage, as inspired by scanning electron microscopy measurements.¹ The ‘folded sheet’ arrangement of fibrils is shown together with the fiber polar angle versus depth assumed by our model.

Collagen exists in 28 different isoforms with, for example, type I dominating in tendon. Cartilage is composed chiefly of fine type II collagen fibrils (sub-micron-diameter cylindrical structures). In healthy tissues, these fibrils have a well-defined 3D stratified organization into superficial, transitional, and radial zones that are characterized by fiber polar angles varying from 90 degrees to zero degrees as depth increases (see Figure 1).¹ This organization is progressively lost during osteoarthritis. Although these fibrils are generally too small to be visualized directly using light microscopy, their architecture has been extensively studied using destructive techniques such as transmission polarized light microscopy and scanning electron microscopy. A technique that could be used on intact tissue or even in vivo could bring new insights, for example, by determining dynamic changes in response to loading or regional variations in normal patients compared with arthritis sufferers. However, no existing techniques can achieve this. For instance, diffusion-tensor magnetic resonance imaging, which has demonstrated sensitivity to aligned fibers in nerves, suffers from the low rate of diffusion anisotropy in cartilage, which leads to data acquisition times of several hours or longer.²

We are developing a novel optical technique for non-destructively mapping 3D collagen organization in cartilage using polarization-sensitive optical coherence tomography (PS-OCT). This is an emerging variant of OCT that possesses sensitivity not just to backscatter but also to birefringence. With a typical depth penetration of up to 1mm, a depth resolution of 2–20 microns, and real-time 2D cross-sectional imaging speeds, PS-OCT has the potential to determine collagen structure in biological tissues.

We have developed a model to describe the PS-OCT signal obtained by illuminating the cartilage surface at multiple angles of incidence.³ Since birefringence is related to physical anisotropy, the apparent birefringence obtained for different incident beam directions varies with the 3D orientation of the fibers. To model the expected signals we have used theory originally developed for the treatment of liquid crystal displays: extended Jones matrix calculus (EJMC). This model calculates the expected retardance signals versus depth for different incident beam directions, given an assumed axial distribution of collagen fiber polar angles. By describing this axial distribution in terms of free parameters (such as the zone thicknesses and polynomial descriptions of the polar angles versus depth), this model can also be used with iterative optimization algorithms to predict these parameters from the multi-angle data sets.

Figure 2. Measurements taken by multi-angle polarization-sensitive optical coherence tomography on cartilage from the bovine fetlock joint (dots) compared with the theoretical model that we have developed (solid line). Measurements at normal incidence and at ±60° in the sagittal plane (plane in which the joint flexes) and coronal plane (the orthogonal plane, also containing the long axes of the bones) are shown.

Figure 2 shows illustrative results obtained using a non-linear optimizer for zone thicknesses, fiber tilt parameters, and so forth. The agreement is good, indicating that the outputs from the optimizer could be used to infer structural properties of the cartilage, such as the superficial zone thickness. Since this parameter potentially changes during the progression of osteoarthritis, its measurement could bring significant benefits to the management of the condition, for example, by revealing the degree of surface zone erosion at various sites on the cartilage surface.

The next phase of this research will be to attempt to make these measurements in human subjects in vivo. To do this, we are investigating novel imaging probes that can provide angle-resolved measurements using a narrow-diameter rigid endoscope. We will then investigate how structural parameters such as fiber orientation vary from site to site across the cartilage surface and how they are modified in the vicinity of osteoarthritic lesions.

This project was supported by Engineering and Physical Sciences Research Council grant EP/F020422/1.