Adipose-derived stem cells (ADSC) based therapies have the potential to treat cartilage disorders such as osteoarthritis in both animals and humans. The current opinion points to secretions from stem cells in the proximity of the diseased joints as key agents in these new regenerative treatments. In this work, biological insights are derived from native fluorescence by using hyperspectral imaging technique. This approach allows to noninvasively monitor the relative levels of individual biochemicals in the tissue, at a cellular level. In this work, we used hyperspectral imaging and unsupervised unmixing to characterize the effects of specific cytokine treatments on different cartilage layers. These layers are composed of various types of collagen, elastin as well as chondrocytes. Strong autofluorescence of the cartilage chip especially from collagen type II presents a significant challenge for the hyperspectral unmixing analysis because the background signal from the collagen tends to dominate the autofluorescence from the chondrocytes. The latter is used for therapy monitoring, and it allows to compare the merits of specific regenerative treatments of cartilage.
The unsupervised hyperspectral unmixing approach developed in this work, Robust Dependent Component Analysis (RoDECA) provides a robust and detailed biochemical information from the examined cells and tissues, with a proper account of intrinsic cellular heterogeneity. With appropriate hardware adjustment and software modification in this cartilage chip model study, it was possible to analyze the biochemical effects of the stem cell based treatments. Most importantly, collagen I and collagen II have been distinguished for the first time in a label- free manner, which gives a new way of observing the regenerative treatments of the articular cartilage. This method can be extended in the future to the analysis of optically thick tissue.