Research On Gall-stone CT Image Segmentation Based On Fuzzy Clustering Alogrithm

In medical images, there are a variety of uncertain and imprecise factors ingalls-tone CT images, thus it is quite difficult to get the accurate segmentationoutcomes of gall-stone CT images. However, fuzzy theory has a good description tothe fuzzy factors, and approximate optimal segmentation outcomes of gall-stone can begotten by using the fuzzy theory. This paper focuses on research on gall-stone CTimages segmentation based on classical fuzzy C-means (FCM) clustering algorithmand its improved algorithms in order to make the gallstone diagnosis clearer and moreaccurate. In this paper, the main work of this paper as follows:Firstly, in chapter1, a brief introduction about the development and significance ofmedical image segmentation is written.Secondly, in Chapter2, clustering algorithm, fuzzy theory and its representativealgorithm—FCM algorithm are each described.Thirdly, in Chapter3Mahalanobis distance is used to instead of the traditional heEuclidean distance to improve FCM algorithm. The new algorithm based Mahalanobisdistance could overcome the faults when data of images with high-dimensionalfeatures (such as gall-stone CT image) are calculated, and the geometric distribution oftheir data could be adaptively adjusted in order to achieve similar data points from thesmall features, and a good clustering results could be obtained. Then the histogramweight and the spatial constrained weight are introduced into the new FCM algorithmbased on Mahalanobis distance. The two of new improved algorithms could acceleratethe convergence speed of FCM algorithm and conquer the fault, which some clusterspecies of minor pixels are covered by some cluster species of major pixels.Fourthly, in Chapter4, the pyramid principle is introduced into fuzzy kernelclustering algorithm (KFCM) to further improve the convergence speed of the FCMclustering.Finally, in Chapter5, summary and outlook are provided.