Research And Application Of Intelligence Gathering Methods Of Early Screening About Cervical Cytology

In the early screening process of the liquid-based cervical cytology, themicroscopic image acquisition is a very important part. Automatic screening efficiencyis affected by the speed of image acquisition. The quality of the captured images have agreat impact on the results of the automatic screening.Therefore, the acquisition systemis necessary to ensure that the collected image quality, but also guarantee a certainacquisition speed. Due to the smaller depth of field at high magnification microscope,the cervical cells image will appear clear stratification phenomenon in the process of theliquid-based cervical cell image acquisition. And the automated microscope systemcurrently on the market, in the absence of system development, can not achievehigh-quality image acquisition requirements of the cervical cells. Therefore, the cervicalcell image intelligence gathering making use of automated microscope system is animportant research topic of cervical cytology screening.Based on early cervical cancer screening system platform of the NanchangXierdaier Medical Technology Co., Ltd in this paper, a kind of intelligence gatheringtheories and methods on the way of guardian mode have been proposed to achieveautomatic continuous liquid-based cervical cells collected image sequences in theThinPrep cervical cell image acquisition mode. The main contents include:1.Establishing the diagnosis modes of cervical cell image sharpness;2.using the thediagnostic mode in resolution intelligence gathering system, wherein the two parts ofthe application of sharpness value and obtaining of the dynamic parameters.(1) The establishment of definition diagnostic mode includes the identification ofimage sharpness parameters, determination of judgment method, determination ofmulti-feature diagnostic model and etc. As we all know, the clarity of the image isdivided into three levels: clear, clearer, not clear. This paper firstly uses unit pixel graygradient of the nucleus edge of the whole image as a parameter to initially determine theimage clarity, then uses clarity of a single nucleus as a secondary basis to judge thewhole image sharpness. Image clarity diagnosis bases on multiple features, buildsmulti-mode diagnostic model to make an accurate classification of Image clarity.(2)Application of image clarity value in automatic acquisition system. In theprocess of automatic image acquisition, image clarity classification is a key step. Forcervical cell images of different specimens, sharpness values are quite different, but within the same specimen, when the image is clear, their best focus approximatelylocate in the same plane, and there is a certain approximation between their imageclarity value. Therefore, within the same specimen, the image clarity value, which isneeded to obtain dynamically, can be clearly determined as judgment parameter for theimage of next field.(3) Dynamic parameter acquisition. The image clarity value is calculated which isused as judgment standard of image definition. The diagnosis model is establishedthrough experiment to determine whether the next field image needs to be focused. Ifnecessary, calculate the image clarity value, and then set it as the new criterion.Therefore, this criterion is intelligent and dynamic.Intelligent capture images of cervical cells can be achieved by the methoddescribed above to control the microscope lens.