Background: An accurate characterization of neural dynamics in mild cognitive impairment
(MCI) is of paramount importance to gain further insights into the underlying neural mechanisms in
Alzheimer’s disease (AD). Nevertheless, there has been relatively little research on brain dynamics in
prodromal AD. As a consequence, its neural substrates remain unclear.
Methods: In the present research, electroencephalographic (EEG) recordings from patients with dementia
due to AD, subjects with MCI due to AD and healthy controls (HC) were analyzed using relative power
(RP) in conventional EEG frequency bands and a novel parameter useful to explore the spatio-temporal
fluctuations of neural dynamics: the spectral flux (SF).
Results: Our results suggest that dementia due to AD is associated with a significant slowing of EEG activity
and several significant alterations in spectral fluctuations at low (i.e. theta) and high (i.e. beta and
gamma) frequency bands compared to HC (p < 0.05). Furthermore, subjects with MCI due to AD exhibited
a specific frequency-dependent pattern of spatio-temporal abnormalities, which can help identify neural
mechanisms involved in cognitive impairment preceding AD. Classification analyses using linear discriminant
analysis with a leave-one-out cross-validation procedure showed that the combination of RP
and within-electrode SF at the beta band was useful to obtain a 77.3 % of accuracy to discriminate between
HC and AD patients. In the case of comparison between HC and MCI subjects, the classification
accuracy reached a value of 79.2 %, combining within-electrode SF at beta and gamma bands. SF has
proven to be a useful measure to obtain an original description of brain dynamics at different stages of AD.
Conclusion: Consequently, SF may contribute to gain a more comprehensive understanding into neural
substrates underlying MCI, as well as to develop potential early AD biomarkers.