|
The early detection of cervical cancer is of great significance in reducing cervical cancer mortality , cervical smear screening technology is now commonly used checks . Smear screening techniques used for manual sheet , however , inefficient , and subject to subjective factors defects thereby reducing the correct rate of screening . Therefore, the research and development of liquid-based cervical cytology computer - aided diagnostic test systems analysis instead of manual read the piece through the machine at the same time to improve testing efficiency and to reduce the dependence of the human , to enhance the correct rate of screening is of great significance . There are a variety of techniques of cervical smear screening , liquid-based cytology technology ( LCT) as an emerging technology producer speed , high quality smear , commonly used for domestic hospitals , represent the future trends. Excellent algorithm as a research object LCT cell image , communicate with hospital specialists and some experimental analysis draw the LCT cell image characteristics of the image ; according to these characteristics , a lot of image segmentation and image segmentation at home and abroad framework , propose an image segmentation task decomposition become asymptotic closed-loop control iterative framework coarse segmentation and fine split in two parts ; , respectively, for coarse segmentation and fine segmentation the split purpose and segmentation characteristics , select the appropriate algorithm , respectively the algorithm corresponding improvement to the specific design of the proposed framework . Specific segmentation algorithm, a large number of experiments , and the experimental results show that split the nucleus of the test cases , the abnormal cases reached 100% correct segmentation rate , thus proving the validity of the segmentation algorithm . Proposed segmentation algorithm , through the analysis of auxiliary test diagnostic system requirements , software architecture design ; communication with hospital specialists and learning similar medical system , the design of the system interface , last by VC and MFC software interface and partition the design and development of the module . However , there is still room for improvement on the results of the time and technical reasons , the proposed algorithm, and the realization of the system for a single picture of the nucleus split and split time , it is necessary to further improve and optimize it , which will this step major work .
|