Automated segmentation and tracking of cells in time-lapse microscopy using watershed and mean shift

In this paper, we present a new method combining watershed and mean shift for segmentation and tracking of cancer cell nuclei in time-lapse fluorescence. First, we apply the watershed algorithm to segment the cells in each frame of the video sequence, including clustered cells. Second, mean shift method is employed to track each cell in its cycle progression. The proposed method can automatically segment and track all cells without any manual initialization. Experimental result shows that our method can detect almost all the touching cells and track them successfully, especially in the case of cell mitosis which is a difficult task using traditional methods such as snake and level set.