Recognition of linear stress fibers based on Hough transform.

Overview

OBJECTIVE: To present an algorithm based on Hough transform for recognition and extraction of linear stress fibers formed on exposure to lysophosphatidic acid (LPA). STUDY DESIGN: A ridge set of head points with lower shoulders is calculated, followed by a thinning process shrinking long, narrow regions to regions of single pixel thickness, then converted into a rectangular map whose value is the number of regional points in the path of a straight line at the angle and intercept determined by two coordinates. The location of the maximum in the map is sought, and the corresponding line with an unlimited length is constructed from the paired coordinates. We removed the line before repeating the process for the next longest straight line, continuing until all lines with reasonable lengths are extracted. RESULTS: Application of the algorithm to the stress fiber images of DOV13 cells stained with Texas red-phalloidin on LPA and AG1478 demonstrates close matches between stress fibers in the original images and linear lines. CONCLUSION: An algorithm for recognition of linear stress fibers formed on exposure to LPA is described and applications to stress fiber images using DOV13 cells with Texas red-phalloidin staining are demonstrated.