Research On Cervical Cytology Assistant Screening System

Cervical smear screening plays an important role in reducing the mortality of cervicalcancer. Prior to the 1980s, all of smear must be approved by the naked eyes using the microscope;but the false-negative rate of the examination results is very high. Therefore, cervical smearscomputer-aided detection system has been developed and sold on the market. However, theseimport equipments are very expensive, many hospitals can not afford to buy such expensiveequipment. Therefore, domestic research and development of Cervical Cytology AssistScreening System (CCASS) has high practical value and extensive market potential.In this paper, the CCASS is constructed by the four major subsystems, including precisemachinery subsystem consisting of microscope platform and related accessories, electroniccontrol circuit subsystem consisting of computer, digital I/O card,stepper motor driver andlimit circuit, image acquisition subsystem consisting of biological microscope, CCD cameras,image acquisition card and software subsystem consisting of image processing module, celltesting module and others.CCASS can process the traditional Pap smear; and can also process liquid-based preparation(TCT). In this paper, platform controlling and image acquisition bottom programs are developedin the CCASS; Pathologist can call these bottom programs by the man-machine interface ofCCASS, thus controlling high-precision scanning microscope platform to move along X and Yaxis of the two-dimensional platform, accessing high-quality, high-resolution magnified image todisplay on the monitor simultaneously, Pathologist can directly diagnose cervical cells indifferent regions of smear on the monitor, which is reducing pathologist's eye fatigue.The software of CCASS is developed on C++ Builder5.0.Basic functions of the softwaresystem include patient information archiving, photo storage, report printing, etc.. In this paper,the cervical cell image is processed targetedly by the image processing algorithms, whichrealizes the cervical cells image geometry transform, color conversion, image smoothing, imageenhancement, edge detection.On the basis of comparative analysis of the relevant micro-imagemosaicing algorithms, a suitable algorithm is choosed to combine local images to ahigh-resolution images of greater vision, which is better for pathologist to observe cervical cellmorphology. By some measurement algorithms of pathological cells and metal particulate which are reported in magazines, cervical cell parameters measuring and nucleus plasm ratio analysisare better realized.Finally, the author has done some researches about popular neural network technologies forthe identification and classification of cervical cells, and has done some simulation work in thepaper.