General method for classification of fiber families in fiber-reinforced materials: application to in-vivo human skin images

Abstract

Fiber structures play a major role for the function of fiber-reinforced materials such as biological tissue. An objective classification of the fiber orientations into fiber families is crucial to understand its mechanical properties. We introduce the Fiber Image Network Evaluation Algorithm (FINE algorithm) to classify and quantify the number of fiber families in scientific images. Each fiber family is characterized by an amplitude, a mean orientation, and a dispersion. A new alignment index giving the averaged fraction of aligned fibers is defined. The FINE algorithm is validated by realistic grayscale Monte-Carlo fiber images. We apply the algorithm to an in-vivo depth scan of second harmonic generation images of dermal collagen in human skin. The derived alignment index exhibits a crossover at a critical depth where two fiber families with a perpendicular orientation around the main tension line arise. This strongly suggests the presence of a transition from the papillary to the reticular dermis. Hence, the FINE algorithm provides a valuable tool for a reliable classification and a meaningful interpretation of in-vivo collagen fiber networks and general fiber reinforced materials.