Temporal Structure in Resting-State fMRI under Marginal and Cross-Spectral Surrogate Controls

Temporal complexity in resting-state fMRI from 270 ADNI participants is not reproduced by IAAFT or cross-spectrum surrogate controls, with strongest-outlier rates of 21.7% (SampEn), 12.7% (MFDFA), and 44.2% (Higuchi).

• Why it matters: Spectral connectivity analyses may miss temporal organization linked to neurological and psychiatric disease, motivating biomarkers based on recurrence, timescale variation, and signal roughness.
• What they did: In 732 ADNI scans across cognitively normal, mild cognitive impairment, and dementia groups, the study computed SampEn, MFDFA, and Higuchi curve shapes within Yeo-7 networks, then compared each real recording to 20 randomized references using network-wise outlier counting.
• The result: Cross-spectrum-preserving references still showed high strongest-outlier rates (39.8%, 17.5%, 49.8%), and although measures were repeatable across visits, they did not improve retrospective prediction of CDR-SB changes.

AI summary of the abstract · may contain errors · not medical advice · read the paper.