To develop a macromolecular-specific MRI contrast, correlation time, as an indicator of the structural changes in the articular cartilage.

Experimental: Human osteochondral specimens (diameter d=6 mm, N=14) from tibial plateau were obtained from patients who underwent a total knee replacement surgery (with the approval of the local ethics committee). MRI was performed at 9.4 T (Oxford instruments Plc, Witney, UK) with a 19-mm quadrature RF volume transceiver (RAPID Biomedical GmbH, Rimpar, Germany) and Varian DirectDrive console (Varian Inc. Palo Alto, CA, USA). Prior to imaging, the specimens were thawed, placed inside a Teflon test tube and immersed in perfluoropolyether, with cartilage surface perpendicular to $B_0$. A magnetization preparation block was modified for continuous wave (CW) $T_{1\rho}$ dispersion measurements, with spin-lock (SL) powers $\gamma B_1$ = 125, 250, 500, 1000 $Hz$,followed by a fast spin echo readout (TR = 5 s, ETL = 4, $TE_{eff}$ = 5 ms, 256 $\times$ 128 matrix size, slice thickness 1 mm, FOV of 16 $\times$ 16 $mm^2$, 62.5 $\mu m$ depth-wise resolution). $T_2$ was measured with spin echo technique. The specimens were stratified using histological OARSI grading ¹ : early osteoarthritis (OA) (grade < 1.5, N = 5) and advanced OA (grade > 1.5, N = 9). Bovine osteochondral cylinders (d = 25 mm, N=6) were drilled from intact patellae, and cut to three separate sections. To induce a specific collagen or proteoglycan (PG) depletion, one section was digested in 30 U/ml collagenase and the other in 0.1 U/ml chondroitinase ABC, respectively. The remaining section was used as an intact control. All sections were incubated at +37ºC for 44 hours and subsequently frozen. Ultimately, a smaller osteochondral cylinder-shaped (d = 7.2 mm) sample was drilled from the center of each section for MRI. MRI was carried out using the same setup as for the human samples. The MRI measurements for human and bovine data are detailed in references ^2,3.

Theory: Correlation time ($\tau_c$) maps were generated by fitting $\tau_c$ from relaxation dispersion measurements of both bovine and human cartilage specimens. The fitting was done for each voxel by using a mathematical model based on the measured on-resonance $R_{1\rho} = \frac{1}{T_{1\rho}}$ relaxation dispersion. The fit function describes the dispersion of the $T_{1\rho}$ spin-lock relaxation time ⁴ where the constants A and B and in particular $\tau_c$ are the fitting parameters: $$R_{1\rho} = \frac{3A \tau_c}{1+4\omega_{SL}^{2}\tau_c^{2}}+B$$ For further analysis, mean depth-wise profiles of cartilage were calculated from the relaxation time and $\tau_c$ maps.

$\tau_c$ maps (Fig. 1) clearly showed the laminar structure in the bovine control specimens. In PG- depleted specimens the laminar appearance was seen as well, but the $\tau_c$ values differed from the controls. In collagen-digested specimens the superficial zone disappeared from the $\tau_c$ maps. In human specimens (Fig. 2), $\tau_c$ maps of early OA showed the trilaminar structure, unlike the advanced OA maps where the cartilage structure appeared disordered. $\tau_c$ profile of bovine control specimens (Fig. 3) and early OA human specimens (Fig. 4) demonstrate clear laminar appearance with two peaks representing superficial and deep zones and a valley in between which corresponds to the transitional zone. $\tau_c$ mean profile of the collagen-digested (Fig. 3) and advanced OA (Fig. 4) specimens revealed a lower $\tau_c$ values in the superficial zone. In contrast to $\tau_c$ profiles, $T_2$ profiles contained one peak representing the transitional zone.$\tau_c$ is a physical property of the tissue describing the dynamics of the water molecules close to the macromolecular constituents of the tissue. In the present study, $\tau_c$ maps showed cartilage tissue structure and degenerative changes. Compared to $T_2$ and $T_{1\rho}$, $\tau_c$ mean profiles are more likely to reveal the degradation in the superficial zone. Despite of the informative images, the results noticeably contained salt-and-pepper noise that represents the over- and under-fitting points. A 3×3 median filter cleared the noise; however, utilizing larger spin-lock frequency range should prevent most of the noise.$\tau_c$ maps, which are fitted from $T_{1\rho}$ relaxation dispersion maps, represent a fundamental physical property of the cartilage tissue. These findings propose that $\tau_c$ can be a novel MRI contrast due to the information it contains about the cartilage structure and its changes.