Objective: Applying graph theory, we investigated how cortical sources small worldness (SW) of resting EEG in eyes-closed/open (EC/EO) differs in amnestic mild cognitive impairment (aMCI) and Alzheimer's disease (AD) subjects with respect to normal elderly (Nold).
Methods: EEG was recorded in 30 Nold, 30 aMCI, and 30 AD during EC and EO. Undirected and weighted cortical brain network was built to evaluate graph core measures. eLORETA lagged linear connectivity was used to weight the network.
Results: In Nold, in EO condition, the brain network is characterized by more SW (higher SW) in alpha bands and less SW (lower SW) in beta2 and gamma bands. In aMCI, SW has the same trend, except for delta and theta bands where the network shows less small worldness. AD shows a similar trend of Nold, but with less fluctuations between EO/EC conditions. Furthermore, in both conditions, aMCI SW architecture presents midway properties between AD and Nold. At low frequencies (delta e theta bands) in EC, aMCI group presents network's architecture similar to Nold, while in EO aMCI, SW is superimposable to AD ones.
Conclusions: In resting state condition, aMCI small-world architecture presents midway topological properties between AD subjects and healthy controls, confirming the hypothesis that aMCI is an intermediate step along the disease progression.
Significance: We proposed the application of graph theory to EEG in reactivity to EO in order to find a marker of diagnosis that - in association with other techniques of neuroimaging - could be sensitive to the progression of MCI or conversion into AD.
Keywords: EEG; Functional connectivity; Graph theory; Resting state; Small-world networks; eLORETA.
Copyright © 2015 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.