Spatial-temporal graph convolutional network for Alzheimer classification based on brain functional connectivity...

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Diagnostics

Background:

• Functional brain connectivity is crucial for understanding brain networks and mechanisms.
• Early diagnosis of Alzheimer's disease (AD) remains a challenge.
• Electroencephalogram (EEG) offers potential for non-invasive brain monitoring.

Purpose of the Study:

• To develop a novel deep learning model for improved early diagnosis of Alzheimer's disease (AD).
• To leverage both spatial-temporal functional connectivity and EEG signal dynamics for enhanced classification.
• To investigate the impact of normal aging on brain network characteristics.

Main Methods:

• Introduction of a dynamical spatial-temporal graph convolutional neural network (ST-GCN).
• Simultaneous consideration of functional connectivity adjacency matrix and EEG signal dynamics.
• Application of 1D convolution for capturing discriminative dynamic temporal information.

Main Results:

• The proposed ST-GCN achieved a classification performance of 92.3% on clinical EEG data.
• The method outperformed existing state-of-the-art approaches for AD diagnosis.
• Demonstrated the model's ability to differentiate between AD patients and healthy controls.

Conclusions:

• The novel ST-GCN model shows significant promise for the early and accurate diagnosis of Alzheimer's disease using EEG.
• This approach enhances understanding of brain network alterations associated with normal aging.
• The findings support the utility of advanced AI techniques in neurological disorder diagnostics.