Automatic diagnosis supporting system for cervical cancer using image processing

In this paper, we suggest methods of automated screening system for cervical cancer to support cyto-pathologists because there is the lack of number of pathologists. It is important to extract nuclei accurately for the automatic diagnosis supporting system. Cells firstly need to be divided into the blue cells and red ones because cells are dyed in different colors based on type of cell: superficial, intermediate squamous and basal cells. Nuclei are then extracted by superpixel segmentation. Finally, we detect malignant cells by nuclear enlargement and color density of nuclei, and distinguish between positive images and negative ones.