Research On The Extraction Of Lymphoma Cells Based On Image Segmentation

Malignant lymphoma is also called "lymphoma", which is one of China’s ten common malignant tumors. This disease often occurs in young, has become a serious threat to human life. At present, the medical staff mostly judgment about the illness of patients by observing the patient’s tumor biopsies. But because in the process of production, the section suffers the influence of external conditions such as equipments and lights, it’s easy to cause slice image ambiguous and difficult to distinguish. At the same time manual interpretation will also bring error, affect the final condition analysis. Therefore, through the cell microscopic image processing as soon as possible to lymphatic tumor cell diagnosis and identification, which is of great significance to save lives. Automatic segmentation and recognition of tumor cells has become the hot spot of research.In order to analyze the recognition of tumor cells quickly and accurately, we need to collect some characteristic dates of tumor cells, to extract the features of cell morphology. Thus the technology of cell division becomes a key. In this paper, According to the clinical Lymphatic cancer cell slice images’characteristics, this paper presents an automatic segmentation method of tumor cells. The main works are as follows:First of all, this paper describes the development of the situation at home and abroad for cell segmentation, and through the traditional threshold segmentation, edge detection and region segmentation of three kinds of segmentation methods to deal with Lymphatic cancer cell image respectively. Through the experimental results, analysis the advantages and disadvantages of each algorithm.Secondly, this paper put forward the extraction method of cancer cells. Through the analysis the characteristics of lymphatic cancer cell image, using K clustering on the luminance component in HSI space, locating the position of cell, then combining with the saturation component cells surrounding the point for further processing, improving the cell edge defect problems in crude extract. Finally complete extraction of target point. Third, the adherent cell segmentation, this paper presents an improved method for the adherent cell concave point locating. By drawing the boundary marker map of cells, the two-dimensional problem of cell boundary is transformed into1d curve problem of distance and Angle, which belongs to Boundary point and the center point. Define the shortest distance for pits. Connected the two pits, realize effective segmentation of cell adhesion.Finally, extract of geometric parameters of cells, including the perimeter, area, circularity, long and short axis of cells, and counted the number of cells, and lay the foundation for cell recognition later.