Semi-automated cell counting in phase contrast images of epithelial monolayers

Non-destructive, quantitative evaluation of cell numbers in epithelial monolayers is necessary to monitor their proliferative rate in a single culture over a time period. Imaging using phase contrast microscopy is non-invasive and creates contrast in optically transparent cells; however intrinsic image artefacts mean that thresholding-based segmentation methods alone are insufficient to identify cells. The method described uses mathematical morphology to assign contrast to cells in a way that enables approximate segmentation by applying a threshold value. Kmeans clustering is subsequently applied to improve enumeration accuracy by removing segmentation artefacts. Cell numbers in images over multiple time-points may then be used to non-destructively generate growth curves. To validate this method we demonstrate a dose dependent relationship between H400 keratinocyte proliferation and concentration of foetal calf serum.