The ability of articular cartilage to withstand cyclic loading during activities of daily living is dependent on a set of complex interactions between the solid extra-cellular matrix and the fluid that exits upon compression and re-imbibes during recovery (Kwan et al., 1984, Linn and Sokoloff, 1965, Mow and Mansour, 1977). Depending on the frequency and severity of the loading, articular cartilage becomes unable to equilibrate between loading cycles and this results in the progressive accumulation of persistent strain, which aligns with the materials science concept of ratcheting (Kang, 2008, Miller et al., 2016, Zhu, 2018). Distinct waveforms with varied recovery time have a clear effect on chondrocyte biology. In contrast to the anabolic effect of dynamic loading, static loading and excessive force have been linked to catabolic gene expression (Guilak et al., 1994, Jones et al., 1982, Sah et al., 1989). Given the importance of tissue mechanics to the biological response, dissecting the intricacies of cartilage recovery may lead to a better understanding of the initiation and progression of joint degeneration that occurs during osteoarthritis. One challenge in cartilage mechanics is the high degree of complexity associated with modeling tissue motion under load due to high fluid content (Torzilli et al., 1990), anisotropic distribution of cartilage constituents (Muir et al., 1970, Xia et al., 2008) and tension–compression nonlinearity (Huang et al., 2003, Korhonen and Jurvelin, 2010, Soltz and Ateshian, 2000) (see Table 1).

Biomechanical testing platforms have evolved over time as theory and technology matured (Cohen et al., 1998, Komeili et al., 2018, Lee et al., 2023, Linn and Sokoloff, 1965, Oloyede et al., 1992, Sah et al., 1989, Torzilli et al., 1997, Zhang et al., 2022b). A common trait linking these systems is direct and constant contact of the platen with the sample throughout the loading cycle. While constant contact with the sample allows for a straightforward approach to measure the height and strain of a sample, such designs place a pre-tare on the sample that may interfere with fluid recovery in cartilage. Fluid entry takes place during the unloading phase of a compression cycle due to negative pressure gradients (Park et al., 2003, Suh et al., 1995), which forcibly draws water back into the tissue. The superficial zone of cartilage has higher permeability than deeper zones (Maroudas and Bullough, 1968), and previous experiments have shown superior fluid rehydration when the top surface is unobstructed (Torzilli et al., 1983, Torzilli and Allen, 2022). Accurate accounting of interstitial fluid in the tissue is important for the study of cartilage under cyclical compression, as fluid pressurization has a prominent role in load support (Krishnan et al., 2005) and excessive water loss facilitates strain accumulation (Maroudas, 1979).

Maintaining contact with the sample while applying loading necessitates a small pre-tare and precludes unobstructed fluid re-entry in between loading cycles. Further, the assumption for measuring strain recovery during unloading is that the platen and sample surface remain coincident, but this can be violated at frequencies as low as 0.1 Hz (Mow et al., 1990, Suh, 1996). Another approach is to employ closed-loop force feedback systems capable of adjusting platen position to ensure contact. However, even fast response times of one hundred milliseconds (Stolberg-Stolberg et al., 2018) would be insufficient to ensure contact in dynamic loading situations. Barker and Seedhom did incorporate lift-off in a compression device using a cam and follower mechanism (Barker and Seedhom, 1997), but this approach limits loading profiles to very simple sinusoidal shapes.

In addition to challenges with maintaining the intended contact during unloading phases, utilizing the platen position to measure strain only provides information in the axial direction, and an orthogonal system for assessing lateral strain would be required. Studies that investigate lateral strain using imaging modalities such as MRI (Eckstein et al., 1999, Lee et al., 2023, Neu et al., 2005) or bright field microscopy (Gao et al., 2019, Komeili et al., 2018) have largely focused on static loading. The lack of lateral strain information during dynamic loading is a missed opportunity to assess the contribution of the collagen fibers in tension, which provide flow-independent viscoelasticity to the tissue response (Mak, 1986, Mononen et al., 2012).

Motivated by the points above, we developed a custom-built device that allows for unobstructed imbibing from the fluid bath during the recovery phases of cyclical compression. The integrated camera captures two-dimensional profiles at a temporal resolution of 30 Hz. The difficulties of converting images into displacement information were circumvented by using deep learning assisted semantic segmentation to provide axial and lateral strain values. We used these capabilities to explore the importance of fluid re-entry during dynamic loading. Our hypothesis is that cyclical loading with a truncated unloading phase will have higher rates of strain accumulation. We tested this hypothesis by measuring strain accumulation in the axial and lateral directions in response to two dynamic loading patterns that differ in the amount of time given for fluid recovery between loading cycles.