Research And Design Of The Computer System Base On Identifying The Neck Cancer Cell

Early cervical smear screening plays an important role in improving the rates of diagnosing cervical cancer and reducing the mortality rate of cervical cancer. But the naked eye detection goes with high false negative rate. Using computer-aided cytology screening system to diagnose cervical cancer can greatly improve the inspection efficiency and reduce human error.Cervical smear automation research and development is divided into three directions: image processing technology is used to achieve some or all of automation, AutoPap as representatives; Pap smear technology is used and automation, ThinPrep as representatives; the semi-automatic detection based on artificial intelligence Interactive, PapNet as the representative. PapNet select the variated images that would be ignored through automated pattern recognition, which enhance the degree of sensitivity, but there is a longer processing time, the higher cost of inspection, and other issues. This article follows the direction of an interactive artificial intelligence and neural network technology so that analyzing the performance of the different cells in the smear with the dyeing, thickness, and the overlapping patterns is possible, and strives to improve efficiency and the correct recognition rate of screening system.This software is developed on the Visual C + +6.0 platform . This software is a typical pattern recognition system, including the task of pattern recognition in many areas: image pre-processing, image segmentation, feature extraction, classification, and so on. In this paper, TBS classification screening is used and this method is superior to the traditional Pakistan-grade law in the reduction of false negative and accuracy rate . Segmentation threshold is used to improve the efficiency of the image segmentation. The method of 8 chain code is used to achieve the tracking of outline of the image and count automatically, this method is more effective in extracting the morphology of the cell. BP three-tier artificial neural network is used to design classification model, momentum - Adaptive learning rate adjustment algorithm is used to improve BP algorithm.The system has achieved the color conversion of the cervical cells image, image enhancement, image segmentation, measurement of cells characteristic parameters and classification, and other functions, but failed to segment and identify overlap cervical cancer cell; segmentation algorithm of cell image, extraction algorithms of cell feature and recognition algorithm still needs to be further optimized.