Brain dysfunction assessment in Alzheimer's disease: A phase-space projection and interactive signal decomposition...

This study introduces and evaluates a signal processing framework to identify electrophysiological biomarkers for neurodegenerative diseases from resting-state electroencephalography (EEG) data. The objective is to quantify distinct patterns of brain dysfunction in Alzheimer's Disease (AD) by analyzing nonlinear signal dynamics and functional connectivity. The analysis was performed on a dataset of 65 participants, including individuals with AD, Frontotemporal Dementia (FTD), and Healthy Controls (HC). A two-stage analytical method was applied to the EEG signals. The first stage, Phase-Space Projection (PSP), transforms the EEG time-series into a signal representing its deviation from a stable reference state. The second stage, Interactive Signal Decomposition (ISD), separates this deviation signal into oscillatory components and a nonlinear residual. Functional connectivity was assessed using the directed Phase Lag Index (dPLI). The AD group exhibited significantly lower functional connectivity compared to the HC group. Concurrently, the mean residual energy, a measure of signal complexity derived from the ISD framework, was significantly lower in AD patients. A direct positive correlation was found between residual energy and Mini-Mental State Examination (MMSE) scores, linking reduced nonlinear signal complexity to greater cognitive impairment. The combined PSP-ISD framework provides a quantitative measure of local nonlinear signal complexity, which is reduced in Alzheimer's Disease. When used alongside functional connectivity analysis, this approach offers a method for characterizing the neuropathology of AD through both network-level degradation and local reduced nonlinear signal complexity, supplying a new analytical tool for clinical neuroscience.