Abstract
In evaluation of image, the results provided by image processing and statistical processing are more objective than those which are judged by visual observation. We propose a method to create a binary image of cell body region from a local area image of Podocyte. The structure of Podocyte is changed by disease. In order to capture this change, we use electron microscope image of Podocyte and select five kinds of mice; wild type, mouse with diabetes mellitus, aged mouse, knockin mouse and knockin mouse with diabetes mellitus. We use contrast adjustment and opening of mathematical morphology for preprocessing to define outline of a cell. And for getting the binary image, we adopt Sauvola method which is a kind of dynamic thresholding. After that we calculate area and circularity of cell body regions to examine difference among five types of mice. From this calculation we can find that the features of wild type mice get near to those of knockin mice as they become diabetic or old. This study includes trial-and error procedure when we should decide some parameters in image processing. To decide these suitable parameters by calculation is a future task.
