The Research And Application Of Scale Invariance In Cervical Cancer Cytology Screening

Early detection of cervical cancer can largely reduce the mortality rate of cervical cancer. So the development of early cervical cancer cytology screening system, is of great significance.This article is based on the images of cervical cell specimens, and use the cervical cancer screening system platform of the Nanchang Cyto Diagnosis S&T Co., Ltd. to collect the cell images. To shorten the time used for the detection and to ensure the detection efficiency of the system, subsample of the specimen image is a effective preparation process. But at the same time, we must ensure that the image content remained unchanged after the sampling, so the selection of sampling ratio is very important. The Combination of image scaling invariance and a multi-mode process will provide a reliable guarantee for the detection of cervical cytology screening system and greatly improve the detection efficiency.This research focuses on the combination of visual invariance and cell pathology,and describes the invariance of cervical specimens contents, and classifies the images in the optimal sampling ratio. The specific work is as following:(1)Combined with the theory of the human visual system and visual invariance,this paper describes the use of scale invariance in perception of the inflammatory cell concentration criterion. It also describes the cervical cells images of inflammatory cell coverage from the perspective of existence of inflammation and the severity of inflammation, then perform the classification criterion of images.(2) Described the lump coverage of specimens. This paper mainly introduces the classification of massive objects in images, and uses the visual invariance to describe the human eyes' invariant perception of the lump in the scaled images. Finally use experiments to verify whether the experimental result is favorable to human visual perception, and give the multi-mode criterion for the lump coverage of specimens,which improves the assessment accuracy and reliability.(3) Described the satisfaction to the cell number in the specimen image, then use multimodal processing methods to analyze the cells number of the scaled cervical specimens. Then determine the satisfaction of cells number in specimens.This paper introduces the theory of the human visual invariance and describes the scaled cervical specimens' contents, describes the image contents from the visualangle, verifies the results of human perception are same with intuitive data, and provide a effective method to improve the detection efficiency for cervical cancer screening system.