Research On The Macrophage Image Segmentation Algorithms Based On Morphology And Region Merging

Macrophages, as a kind of immune cells in the body, has the function of anti-tumor and immune regulation, they can swallow the apoptotic cells and the bacteria invading the body, forming the phagosome. Macrophages play an important role in wound healing, enhancing the body’s immune ability, disease prevention and treatment. Due to the macrophages have the characteristic of different size, irregular shape and strong adhesion phenomenon, so realizing macrophages automatic detection is more difficult than red blood cells and white cells detection, the main difficulty is correctly segmenting the images of adhesion macrophages.In this paper, we are focused on the research of segmentation algorithm to adhesion macrophage image. With the morphology being used in image segmentation, watershed becomes one of the most efficient skills in solving adhesion cells. However, the traditional watershed algorithm is apt to cause over-segmentation for the noise and irregular shape of macrophages. If extracting the "seeding points" from the distance map, and optimizing "seeding points" depending on the location and grayscale information of "seeding points". Then redistributing distance map based on the optimized "seeding points" before using watershed algorithm. It can efficiently segment adhesion cells as well as restrain the over-segmentation. In addition, we propose a novel algorithm combining watershed and region merging, counting the similarity between each two sub-region according to the characteristic parameter of the regions, it not only can restrain over-segmentation phenomenon, but also can locate the edge of target more accurately. In order to analyse the inner circumstance of the cells, we use the adaptive multi-scale morphological gradient to extract the cell image contour line. After segmentation, we can mark the numbers next to the cells and abstract the characteristic parameters such as circumference, area, roundness factor, mean gray, secondary moment and so on, providing data support to macrophage research and clinical diagnosis.