Analysis of cell behavior on micropatterned surfaces by image processing algorithms

With the recent developments in medicine and biology experiments a large amount of data is gathered in the form of multimedia elements (images, videos). Many algorithms have been developed and adapted based on the system of interest, and often the most challenging feature of the images may be used to facilitate a better analysis of the image. Herein, we developed an image analysis algorithm for quantification of cellular shape and size on micropatterned surfaces (gelatin) as a means to predict their phenotypic behavior. We have two conditions: i) individual dispersal of the cells on the surfaces, and ii) the clustering of cells in small and large patches. The analysis of the former condition, that includes counting and determining of the cells' area, relies on successful segmentation. In the second setting where clustering of cells is favoured, individual cell segmentation and counting becomes more challenging and we determine the relative area that is covered by the cells. Direct image processing techniques can provide a reasonable qualitative picture of the behavior of the cells that sit on the regularly micropatterned surfaces that create a challenging background for the segmentation. Employing filters in both spatial and frequency (reciprocal) domain enabled a better quantitative analysis of the cell behavior. Our method uses the periodic repetition of the patterns to distinguish the cellular features from the topography of the substrate, which can be generalized for the analysis of cellular metrics on micropatterned surfaces.