Phase contrast images of keratocytes were converted into binary images using custom-written Matlab codes. Briefly, we used Matlab edge detection and a basic morphology function to outline cells in the phase contrast image. We used the Otsu method (Otsu, 1979 ) to erase the halo artifacts. If shape was still unsatisfactory, we then used the Lasso tool in Photoshop (Adobe) to manually draw the cell shape. Polygonal outlines extracted from the binary images were plotted in Celltool, an open source software (Pincus and Theriot, 2007 (link)). Geometric features of each cell, including centroid, area and aspect ratio, were measured directly from the polygons using standard formulas (Barnhart et al, 2011 (link)). The aspect ratio was calculated as the ratio of the width to the length of a cell and was used to determine if a cell was polarized or un-polarized. To define polarization, we conducted a pilot experiment and traced a large number of cells undergoing spontaneous polarization. The significant polarization was detected around 15 minutes. We then calculated the standard deviation of the aspect ratio at this time point (214 cells were analyzed). A value out of range of mean ± SD (0.8547 to 1.1687) is treated as polarized (Supplemental Figure 1).