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 ability of the “novelty indices” (NI) kurtosis, negentropy, and 1-lag autocorrelation, to discriminate between potentially interesting (“structured”) and uninteresting (“noisy”) time-courses (TCs). NIs may be employed as preprocessing tools to focus only on the interesting TCs prior to any further exploratory or confirmatory approach.