Accurate Extraction of Contour Information from the Cleavage of Neural Stem Cells

Accurate extraction of contour information from neural stem cells is important, as the shapes of neural stem cells may contain significant features for studying their activities in time lapse image sequences especially during their cleavage. In this paper, an algorithm is presented for this purpose. Fuzzy threshold method based on Zadeh´s maximum entropy is adopted for cell segmentation. The best parameters of maximum fuzzy entropy are computed using genetic algorithm. The acquired contours are then smoothed in their one-dimensional Centroid-Contour Distance (CCD) diagrams. Compared with the results derived from traditional segmentation methods, the contours extracted using our method are much closer to the real cells, even for tiny objects size down to 30 pixels in images.