Analysis of Electroencephalographic Dynamic Functional Connectivity in Alzheimer’s Disease

Abstract

The aim of this study was to characterize the dynamic functional connectivity of resting-state electroencephalographic (EEG) activity in Alzheimer’s disease (AD). The magnitude squared coherence (MSCOH) of 50 patients with dementia due to AD and 28 cognitively healthy controls was computed. MSCOH was estimated in epochs of 60 s subdivided in overlapping windows of different lengths (1, 2, 3, 5 and 10 s; 50% overlap). The effect of epoch length was tested on MSCOH and it was found that MSCOH stabilized at a window length of 3 s. We tested whether the MSCOH fluctuations observed reflected actual changes in functional connectivity by means of surrogate data testing, with the standard deviation of MSCOH chosen as the test statistic. The results showed that the variability of the measure could be due to dynamic functional connectivity. Furthermore, a significant reduction in the dynamic MSCOH connectivity of AD patients compared to controls was found in the delta (0–4 Hz) and beta-1 (13–30 Hz) bands. This indicated that AD patients show lesser variation in neural connectivity during resting state. Finally, a correlation between relative power and standard deviation was found, suggesting that an increase/peak in power spectrum could be a pre-requisite for dynamic functional connectivity in a specific frequency band.