Adaptive marker-based watershed segmentation approach for T cell fluorescence images

It is intractable problem to segment the fluorescence image of T-cells with different sizes, irregular shape, and severe overlapping by conventional marker-based watershed segmentation. In this paper, Adaptive Marker-controlled Watershed method (AMWS) will be proposed. The Otsu strategy firstly is performed twice in a row to capture as many T-cells as possible. Then based on T-cells' roundish shape, an improved strategy to obtain markers adaptively is present using the evaluation of the segmentation result. This strategy is able to mark the single cell and the overlapping cells accurately. It avoids the ineffectiveness of ultimate erosion which is due to different sizes of cells. The experimental results show that the proposed strategy in this paper can effectively avoid over-segmentation and under-segmentation thus improves both accuracy and robustness of the segmentation.