Image analysis of cytoskeletal network dynamics

• Roman Yakobenchuk
• Dmytro Kotsur
• Galiya Sakaeva
• Bernd Hoffmann
• Rudolf Merkel
• Julian Mattes

Cell migration is essential for phenomena such as tissue formation, wound healing and tissue invasion during carcinogenesis. Sub-cellular structures related to cell migration include actin or keratin bundles, focal adhesions or membrane protrusions. The extraction of geometry and motion of these structures from microscopy data can often be a challenging task, especially the extraction of the actin cytoskeleton and of keratin intermediate filaments. Here, we present a method for identifying a single actin stress fiber in the actin network appearing in live cell images. In order to remove noise from actin tagged fluorescence microscopy images and to enhance the brightness and the contrast of the actin network, we propose an image pre-processing routine, which improves the results of further segmentation. A graph representation of the actin cytoskeleton is used in order to transfer the segmentation problem into the domain of graph processing. Therefore, the task of single stress fiber extraction can be solved by means of a specially adapted shortest path search algorithm. Thereby, the assessment of paths is based on geometrical properties of the corresponding actin fibers. The proposed method is efficient and almost fully automatic, only requiring user interaction in a form of two given end-points in the network. The fiber between these two end-points is detected and segmented. We give an overview of the proposed methodology and report on initial evaluation results obtained for videos of NHEK cells with a complex actin cytoskeleton structure.