Abstract:
Abstract Alzheimer’s disease (AD) and Frontotemporal Dementia (FTD) represent
significant global health challenges, characterized by cognitive impairments
and often regarded as disconnection syndromes with unclear pathological mechanisms
and diagnostic complexities. While electroencephalography (EEG) offers a
promising, cost-effective avenue for identifying neurological biomarkers, its analysis
faces challenges such as volume conduction and interrater variability. Previous
network analysis approaches in this domain have often relied on binary networks,
potentially overlooking crucial information in weighted connections, and suboptimal
community detection algorithms like Louvain, which can produce unreliable
results.
This study addresses these limitations by performing a comprehensive network
analysis of AD and FTD using EEG data, with electrodes as nodes and
functional connectivity measures as edges. The methodology involves rigorous
EEG pre-processing, including Butterworth bandpass filtering, Artifact Subspace
Reconstruction (ASR), and Independent Component Analysis (ICA). Data is segmented
into 12.288-second epochs and filtered into five distinct frequency bands
(delta, theta, alpha, beta, gamma). Functional connectivity is computed using
the Phase Lag Index (PLI) for delta and theta bands, and Amplitude Envelope
Correlation with Leakage Correction (AEC-c) for alpha, beta, and gamma bands,
applying specific thresholds. Adjacency matrices are constructed, averaged across
epochs, and then at the group level. Key network parameters (e.g., mean node
degree, clustering coefficient, path length, efficiency) and centrality measures are
calculated. Crucially, the study employs the Leiden algorithm and average-linkage
hierarchical clustering for community detection, overcoming the shortcomings of
prior methods. Furthermore, this research includes the development of an online
EEG analysis system to facilitate user upload, processing, visualization of results (adjacency matrices, network parameters, centrality measures, clusters), and data
export. This work aims to provide deeper insights into the topological architecture
and functional organization of brain networks in AD and FTD, ultimately
contributing to more objective, data-driven diagnostic methods and a practical
tool for clinical and research applications.
