Hierarchical Fast Clustering Method for fMRI Feature Reconstruction

In order to solve the feature reconstruction problem of fMRI time series, hierarchical fast clustering method (HFCM) is proposed. The reconstruction of features can be thought as finding the task-related region of interest (ROI) in the human brain fMRI in order to eliminate information redundary. HFCM takes advantage of optimizing the hierarchical structure and tuning weights of different kind of distances. Comparing with the existing reconstruction methods, e.g. K-means and t-test, HFCM saves more than 62% running time, on condition of ensuring the precision of task-related estimating.