Abstract
This talk presents an integrative framework for the classification of Mild Cognitive Impairment (MCI) and Alzheimer’s Disease (AD) using resting-state EEG. The approach bridges traditional signal analysis and deep learning methodologies, providing a multi-perspective view of dementia-related brain dynamics.
On one hand, the framework employs complexity measures, spectral rhythms, and synchrony features to capture nonlinear dynamics, frequency-specific alterations, and disrupted connectivity patterns. On the other hand, it leverages deep learning architectures applied to both raw EEG time series and spectral image representations, including CNNs, ResNet, EEGNet, and Vision Transformers, to automatically learn hidden markers of cognitive decline.
By integrating classical neuromarker discovery with modern neural architectures, the presentation highlights resting-state EEG as a powerful and practical diagnostic tool, supporting the early detection and staging of dementia and demonstrating how computational neuroscience can advance clinical decision support in Alzheimer’s and MCI research.