Microstate-Specific Cross-Frequency Coupling Networks: A Spatiotemporal Framework for MCI Identification

Area of Science:

• Neuroscience
• Cognitive Science
• Biomarkers

Background:

• Mild cognitive impairment (MCI) poses a significant challenge for early diagnosis.
• Current diagnostic methods may not fully capture the subtle neurophysiological changes associated with MCI.
• Electroencephalography (EEG) offers a promising avenue for non-invasive brain activity assessment.

Purpose of the Study:

• To develop a novel spatiotemporal framework integrating EEG microstates and cross-frequency coupling for early MCI identification.
• To investigate microstate-specific alterations in brain network synchronization in individuals with MCI.
• To establish a physiologically interpretable EEG biomarker for MCI diagnosis.

Main Methods:

• Resting-state EEG data from 22 MCI patients and 22 healthy controls.
• Derivation of four canonical EEG microstates (A-D) using topographic clustering.
• Construction of microstate-specific cross-frequency coupling networks across delta, theta, alpha, and beta bands.
• Classification using mRMR feature selection and a CatBoost classifier.

Main Results:

• Significant disruptions in cross-frequency phase synchronization were found in microstates A, C, and D in MCI patients.
• The classification model achieved the highest performance in microstate D, with 84.4% balanced accuracy.
• Key discriminating features involved theta band synchronization in the limbic network and alpha band synchronization in frontoparietal/dorsal attention networks.

Conclusions:

• MCI is characterized by microstate-dependent impairments in cross-frequency synchronization, indicating compromised spatiotemporal integration.
• The proposed microstate-specific cross-frequency coupling network framework serves as a novel EEG biomarker for early MCI detection.
• This approach offers a physiologically interpretable method for enhancing MCI diagnosis.