A Level Set Method with Dynamic Prior for Cell Image Segmentation

To well handle the concerned issues, i.e., blur edges and cell adhesion, of cell segmentation in microscopy images, and inspired by the characteristic of cells, we proposed a novel level set method with dynamic shape prior. Firstly, we place a seed in each cell by using a threshold method on the LBP feature of microscopy images, and these seeds are taken into the initialization of level set function. Then, fractional derivatives are introduced into extracting the strength of cell edges that is utilized to design a new cell-edge indicator for driving the closed curve to cell edges. Finally, we incorporate the automatic initialization and the stopper function into a level set method to evolve the curve and finally realize cell segmentation. The comparison with the state-of-the-art methods shows that our method could improve the segmentation performance on adhesive cells with blur edges, and it does not need to tune too many parameters.