A NEW CORRELATION-BASED FUZZY-LOGIC CLUSTERING-ALGORITHM FOR FMRI

Fuzzy logic clustering algorithms are a new class of processing strate gies for functional MRI (fMRI), In this study, the ability of such met hods to detect brain activation on application of a stimulus task is d emonstrated. An optimization of the selected algorithm with regard to different parameters is proposed. These parameters include (a) those d efining the preprocessing procedure of the data set; (b) the definitio n of the distance between two time courses, considered as p-dimensiona l vectors, where p is the number of sequential images in the fMRI data set; and (c) the number of clusters to be considered. Based on the as sumption that such a clustering algorithm should cluster the pixel tim e courses according to their similarity and not their proximity (in te rms of distance), cross-correlation-based distances are defined. A cle ar mathematical description of the algorithm is proposed, and its conv ergence is proven when similarity measures are used instead of convent ional Euclidean distance. The differences between the membership funct ion given by the algorithm and the probability are clearly exposed. Th e algorithm was tested on artificial data sets, as well as on data set s from six volunteers undergoing stimulation of the primary visual cor tex. The fMRI maps provided by the fuzzy logic algorithm are compared to those achieved by the well established cross-correlation technique.