Mining Interaction Patterns among Brain Regions by Clustering Based Interactive K-Means

Functional Magnetic Resonance Imaging (FMRI) provides study of brain functions. The information content from that is large in volume and complex and data requires effective and efficient data mining techniques. To understand the complex interaction patterns among brain regions novel clustering technique is proposed. Each subject consider as FMRI image histogram. The objective is to assign objects exhibiting a similar intrinsic interaction pattern to common cluster. To formalize this idea, define a cluster by a set of mathematical models describing the cluster-specific interaction patterns. Based on this novel cluster notion, propose interaction K-means (IKM), an efficient algorithm for partitioning clustering. IKM simultaneously clusters the data and discovers the relevant cluster-specific interaction patterns. The results on two real FMRI studies demonstrate the potential of IKM to contribute to a better understanding of normal brain function and the alternations characteristic for psychiatric disorders.