A multi-graph convolutional network method for Alzheimer's disease diagnosis based on multi-frequency EEG data with...

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Diagnostics

Background:

• Alzheimer's disease (AD) diagnosis is challenging due to subtle early brain changes.
• Current methods often overlook spatial connections and multi-frequency EEG data integration.
• Electroencephalography (EEG) is a key tool for studying neurodegenerative diseases.

Purpose of the Study:

• To develop a novel graph-based deep learning model for early Alzheimer's disease diagnosis.
• To integrate both functional and structural connectivity from multi-frequency EEG data.
• To improve the accuracy and comprehensiveness of AD detection.

Main Methods:

• Introduction of a Multi-Frequency EEG data-based Multi-Graph Convolutional Network (MF-MGCN) model.
• Integration of functional and structural connectivity to capture brain region relationships.
• Extraction of differential entropy (DE) features from five EEG frequency bands and aggregation using graph convolutional networks (GCNs).

Main Results:

• The MF-MGCN model achieved 96.15% accuracy in classifying Alzheimer's disease (AD) and healthy controls (HC).
• The model demonstrated a high Area Under the Curve (AUC) of 98.74% for AD and HC classification.
• The developed approach outperformed existing methods in AD diagnosis.

Conclusions:

• The MF-MGCN model shows significant potential as a clinical tool for Alzheimer's disease diagnosis.
• Early detection of AD can be facilitated, enabling timely intervention and personalized treatment.
• This novel approach enhances the understanding of dynamic brain activity across different frequency bands for AD detection.