Study On Analysis And Recognition System Of Microscopic Cell Image

With the wide use of computer technology today, computer image processing and anal-ysis play more and more important role in medical field. Study on analysis and recognitionof microscopic cell image has become forefront topic these days. In clinical diagnosis andmedical research, the application of computer technology on automatic process and analysisof microscopic cell image, help doctors make quick and accurate diagnosis.The original computer-aided diagnosis system can scan and read the sample underthe microscope, which provides platform for automatic diagnosis system. Focusing on thecrucial theories and technologies of analysis and recognition system of microscopic cellimage, a deal of research and experimental analysis is stated in this thesis, the major workis as follows:(1) The thesis studies a variety of image segmentation methods. To solve the problem inuneven and overlapping cell images, the thesis proposes the method that based on thresholdmethod and watershed method, and realize the segmentation.(2) Analysis parameters of the cell nucleuses, and give the computing methods of them.Then study the method of support vector machine, choose RBF as the kernel, and train theclassifier. The experiment shows that the types of the cell nucleuses are well classified.(3) Add the algorithm modules to the original computer-aided diagnosis system,then achieve automatic analysis and identification of cell images. Experiment shows that,analysis and recognition system of microscopic cell image combines the expert knowledgeand the exactness and quickness of computers. It can avoid the disadvantages of manualcytological diagnosis, and provide reliable technical methods for early cancer screen anddiagnosis.