Data mining a functional neuroimaging database for functional segregation in brain regions



AbstractWe describe a specialized neuroinformatic data mining technique in connection with a meta-analytic functional neuroimaging database: We mine for functional segregation within brain regions by identifying journal articles that report brain activations within the regions and clustering the abstract of the articles using non-negative matrix factorization on the bag-of-words matrix. We divide the brain activations reported in the articles according to the cluster assignment and test for difference between the spatial distribution of the sets of activations. Among our findings is that the memory and pain functions are spatially segregated within the cingulate gyrus.
Keywordsdata mining, text mining, neuroimaging, positron emission tomography, magnetic resonance imaging, brain, brain mapping
TypeConference paper [Without referee]
Conference
Editors
Year2006    Month August
PublisherThe Department of Computer Science, University of Copenhagen
AddressCopenhagen, Denmark
ISBN / ISSN0107-8283
Electronic version(s)[pdf]
Publication linkhttp://www.diku.dk/publikationer/tekniske.rapporter/rapporter/06-08.pdf
BibTeX data [bibtex]
IMM Group(s)Intelligent Signal Processing