An expert system based on computer vision and statistical modelling to support the analysis of collagen degradation

[Abstract] The poly(DL-lactide-co-glycolide) (PDLGA) copolymers have been specifically designed and performed as biomaterials, taking into account their biodegradability and biocompatibility properties. One of the applications of statistical degradation models in material engineering is the estimation of the materials degradation level and reliability. In some reliability studies, as the present case, it is possible to measure physical degradation (mass loss, water absorbance, pH) depending on time. To this aim, we propose an expert system able to provide support in collagen degradation analysis through computer vision methods and statistical modelling techniques. On this base, the researchers can determine which statistical model describes in a better way the biomaterial behaviour. The expert system was trained and evaluated with a corpus of 63 images (2D photographs obtained by electron microscopy) of human mesenchymal stem cells (CMMh-3A6) cultivated in a laboratory experiment lasting 44 days. The collagen type-1 sponges were arranged in 3 groups of 21 samples (each image was obtained in intervals of 72 hours).