Contribution of Graph Theory Applied to EEG Data Analysis for Alzheimer's Disease Versus Vascular Dementia Diagnosis

Background: Most common progressive brain diseases in the elderly are Alzheimer's disease (AD) and vascular dementia (VaD). They present with relatively similar clinical symptoms of cognitive decline, but the underlying pathophysiological mechanisms are different. Objective: The aim is to explore the brain connectivity differences between AD and VaD patients compared to mild cognitive impairment (MCI) and normal elderly (Nold) subjects applying graph theory, in particular the Small World (SW) analysis. Methods: 274 resting state EEGs were analyzed in 100 AD, 80 MCI, 40 VaD, and 54 Nold subjects. Graph theory analyses were applied to undirected and weighted networks obtained by lagged linear coherence evaluated by eLORETA tool. Results: VaD and AD patients presented more ordered low frequency structure (lower value of SW) than Nold and MCI subjects, and more random organization (higher value of SW) in low and high frequency alpha rhythms. Differences between patients have been found in high frequency alpha rhythms in VaD (higher value of SW) with respect to AD, and in theta band with a trend which is more similar to MCI and Nold than to AD. MCI subjects presented a network organization which is intermediate, in low frequency bands, between Nold and patients. Conclusion: Graph theory applied to EEG data has proved very useful in identifying differences in brain network patterns in subjects with dementia, proving to be a valid tool for differential diagnosis. Future studies will aim to validate this method to diagnose especially in the early stages of the disease and at single subject level.