Novelty indices: identifiers of potentially interesting time-courses in functional MRI data

In a systematic study of hybrid MR time-series with simulated "activation" (contrast-to-noise range 1-10) we investigated the power and compared the a bility of the "novelty indices" (NI) kurtosis, negentropy, and 1-lag autoco rrelation, to discriminate between potentially interesting ("structured") a nd uninteresting ("noisy") tirne-courses (TCs). NIs may he employed as prep rocessing tools to focus only on the interesting TCs prior to any further e xploratory or confirmatory approach. (C) 2000 Elsevier science Inc. All rig hts reserved.