Research Of Recognition And Localization On Cells And Trabeculae Bone In Pathological Images Of Bone Marrow

With the rapid development of computer technology and the combination of computer science and other scientific areas, medicine image processing technology has become an important tool in computer-aid diagnosis. It has a broad application prospect in many areas.In marrow pathological images, the localization and quantitative analysis of immature precursor cells are significantly important for the research of early leukemia diagnosis. However, because of the high complexity of the medicine images, current methods of medicine image processing cannot achieve a desired result in recognition and localization of immature precursor cells and bone trabeculae in marrow pathological images.Based on the above problems, this dissertation studies the localization, recognition and quantitative analysis of immature precursor cells and bone trabeculae in marrow pathological images according to the characters of the research objects and current image processing technology. There are three main parts in this paper:1. Detect the immature precursor cells in marrow pathological images with morphological operations and the characters of the cells themselves. Firstly, use modified OTSU to binarize the image; then process the image with morphological operations, and localize the cells with the sizes of cells.2. Identify the area of bone trabeculae with their sizes and thresholds.3. Based on the above results, calculate the distance between cells and bone trabeculae with distance transformation, divide the area around bone trabeculae into several parts and then quantitatively analyze the cells in different parts.The result in this dissertation is verified through a series of experiments and proved to be effective, which lays a foundation for quantitative and statistical analysis of relation between the localization of abnormal immature precursor cells in marrow pathological slices and leukemia recrudescence.