Fitting of random tessellation models to keratin filament networks.

The role of specific structural patterns in keratin filament networks for regulating biophysical properties of epithelial cells is poorly understood. This is at least partially due to a lack of methods for the analysis of filament network morphology. We have previously developed a statistical approach to the analysis of keratin filament networks imaged by scanning electron microscopy. The segmentation of images in this study resulted in graph structures, i.e. tessellations, whose structural characteristics are now further investigated by iteratively fitting geometrical statistical models. An optimal model as well as corresponding optimal parameters are detected from a given set of possible random tessellation models, i.e. Poisson-Line tessellations (PLT), Poisson-Voronoi tessellations (PVT) and Poisson-Delaunay tessellations (PDT). Using this method, we investigated the remodeling of keratin filament networks in pancreatic cancer cells in response to transforming growth factor alpha (TGFalpha), which is involved in pancreatic cancer progression. The results indicate that the fitting of random tessellation models represents a suitable method for the description of complex filament networks.