Segmentation For Medical Sequence Images

With the development of the medical imaging, medical image processing becomes more and more important. As the key technique of medical image processing, medical image segmentation plays an important role in clinic and pathology research. Quality of image segmentation directly affects the following result of image processing and image segmentation is regarded as the bottleneck of medical image processing.This paper is mainly about the methodologies of sequence image segmentation. There are two ways for sequence image segmentation: one is directly segmenting in 3D data space and extracting the interested object region; the other is segmentation for each slice and composing the object region of all slices.Firstly, the paper briefly introduced the represent methods of medical image segmentation and analysed their characteristics and problems. Secondly, a detailed discussion is given about the active contour models. Then we selected the GVF Snake model for segmentation of pupil sequence images. Thirdly, the paper focused on the methods of graph cuts, introduced the basic theory and the studying situation of graph cuts. Then the paper finished the min-cut/max-flow algorithm and used it for segmentation of CT sequence images.Lastly, the paper studied the weighted aggregation algorithm based on graph. Combining the watershed algorithm and the weighted aggregation algorithm, a fast algorithm for segmentation of sequence images based on region relativity is proposed. As the fast segmentation and low computational complexity of watershed method, we made it as the pretreatment of the weighted aggregation algorithm. The new method solved the lower speed problem for segmentation of big images. Using the statistical information and region relativity, the accuracy of segmentation is improved.