Immunohistochemical Cell Microscopic Image Segmentation Algorithm Research And Application

As a malignant tumor with high death rate, breast cancer has seriously impact onwomen’s physical and mental health, even threaten their lives. In some areas of ourcountry, the incidence of breast cancer is very high. The early detection of breast cancerplays an important role in increasing the cure rate of breast cancer. With thedevelopment of pathology, immunohistochemistry(IHC) techniques have been widelyused in the diagnosis of breast cancer. Through the IHC process after tissue slicing, wecan observe that the lesion areas(the positive area) and the normal areas(the negativearea) of a specimen show a different color response respectively under the microscope.The main purpose of this research is to finish the recognition of positive cells andcompute the rate of positive cells in this slide from the tissue samples which wereprocessed by IHC techniques. The images which were captured by camera under themicroscope transfer to computer, then process and analyze these images. This paperpresents a novel breast cancer detection algorithm based on an improved watershedalgorithm and concave detection. And at last, we can achieve the numbers of positivecells automatically. Besides this paper realize the assess of degree of positive stainingcells based on the theory of Integral Optical Density(IOD). We finished thedevelopment of a Pathology Images Processing system which can help pathologydoctors or the researchers finish the diagnosis in clinical application and research. Themicroscope cells images of breast cancer is the main object of this research, we focus onthe recognition, segmentation, counting and mark of positive cells. The main studyachievements and works of this research are summarized as following:(1) Do research and study on segmentation algorithms of microscope cell image.This paper presents a novel segmentation algorithm process of positive cells. Thisalgorithm can realize the segmentation of cluster cells without losing geometryproperties of cells and finish the recognition, counting and mark of positive cells.(2) Do research on the analyze techniques of IHC microscope cell image. Thedoctor can distinguish the lesions of the breast cancer cells based on the theory ofPositive level index(PLI).PLI is equal to the rate of positive cells*positive intensity. This paper finish the compute of the rate of positive cells based on the recognition andcounting of cells.In addition, this paper realize the assessing of the degree of positivestain areas.(3) Build an image process platform. We develop an image process system withpowerful image process functions and expansion features which can assist to completethe clinical diagnosis. The statistical data can support the final diagnosis conclusion.Besides, the establishment of this platform also laid a foundation for further researchand application.From the view of the last applied effect, the achievements of this research and theapplication of this pathology image processing system play an important role inimproving the efficiency and accuracy of the clinical diagnosis as well as in teachingand research.