Analyzing orientation patterns of human skin fibroblasts using image gradient information.

Fibroblasts are the cells within the dermis layer of skin which are responsible for generating connective tissue, where collagen is the main component. However, the effects of collagen on the motion behavior of human skin fibroblasts were not studied well. We therefore propose an image analysis method to quantify the effects of collagen on the motion characteristics of fibroblasts. Images are divided into disjoint blocks and three gradient based descriptors, namely histogram of oriented gradients (HOG), the modified HOG (modiHOG), and the gradient direction histogram (GDH) are then applied to each block. These features can extract the orientation and shape or orientation only information of the skin fibroblasts from images. An orientation similarity function and a quantitative index, called first minimum similarity distance, are defined to quantify the orientation behavior of fibroblasts. An experiment with two sets of fibroblasts images, obtained from cell culture with and without collagen coating, demonstrates that the skin fibroblasts produce smaller orientation patterns, when they are cultured in the presence of collagen proteins. A statistical t-test is then performed, which validates the significant differences in the orientation behavior of the skin fibroblasts in the two cases mentioned.