Cluster Study Based On Functional Magnetic Resonance Imaging Data

With the gradual development of world economy and technology, the clustering analysis has been widely used in many areas, especially in medical treatment area. Due to the modernization of medical apparatus, it is more and more convenient for people to get data. The research of data deepens as well. In order to get useful information from these complex data, we have to find an efficient method of data analysis. Therefore, high dimensional spatial data clustering analysis becomes more and more important. By cluster method to find the classification principles of the data itself, We are better able to deal with the data.Firstly, the paper gives a detailed introduction of the concerned concepts of clustering, the conditions of clustering algorithms and the two kinds of similarity measurement in the process of clustering. In addition, according to the theory of clustering, we divide clustering into the following kinds:based on hierarchy, based on partitioning, based on fuzzy and so on.Secondly, the paper focuses on the typical clustering algorithm. For example:the systematic clustering, k-means clustering, the fuzzy c-means clustering, etc. The paper mainly expounds the basic concept and thought of systematic clustering, the basic theory of k-means clustering and the theoretical knowledge of fuzzy c-means clustering.Finally, the paper conducts empirical analysis and chooses several different kinds of clustering analysis. For example, the systematic clustering, the k-means algorithm, k-means algorithm based on fuzzy c-means. Through these methods to deal with the human brain resting state functional magnetic resonance imaging data of 62 normal human beings, which is mainly to cluster 90 brain regions of normal human being. Then compare the clustering results with the ALL template brain regions to see whether it is consistent with the fact. Through several different kinds of clustering algorithms, the paper get the optimal clustering results and to make sure whether the 90 normal human being’s brain regions can be distinguished in some degree by clustering. In fact, it turns our to be more difficult to classify human being’s 90 brain regions. But in some certain brain regions, whatever the method you use, it is obviously can be classified in to one category. In other words, if you just conduct rough classification, you can get fuzzy results. But it is more difficult to deepen the classification.