A Novel Method to Identify Mild Cognitive Impairment Using Dynamic Spatio-Temporal Graph Neural Network.

Area of Science:

• Neuroimaging
• Machine Learning
• Computational Neuroscience

Background:

• Resting-state functional magnetic resonance imaging (rs-fMRI) is crucial for mild cognitive impairment (MCI) research, as MCI indicates a higher risk of progression to Alzheimer's disease (AD).
• Existing machine learning and deep learning methods often overlook the crucial spatial and temporal dimensions within rs-fMRI data.

Purpose of the Study:

• To develop and evaluate a novel dynamic spatiotemporal graph neural network model for enhanced analysis of rs-fMRI data.
• To improve the classification accuracy of Alzheimer's disease (AD), mild cognitive impairment (MCI), and normal cognition (NC) using a comprehensive approach to rs-fMRI data.

Main Methods:

• A dynamic spatiotemporal graph neural network model was constructed, incorporating temporal, spatial, and graph pooling blocks.
• The model was designed to extract both the Blood-Oxygen-Level-Dependent (BOLD) signal and the spatial functional connectivity structures from rs-fMRI data.
• The model's performance was evaluated on classifying subjects across three groups: AD, MCI, and normal cognition (NC).

Main Results:

• The proposed dynamic spatiotemporal graph neural network model achieved a classification accuracy of 83.78% for AD, MCI, and NC.
• This accuracy surpasses previously reported results, demonstrating the model's effectiveness in learning from complex spatiotemporal patterns in rs-fMRI data.
• The study highlights the significant role of dynamic spatio-temporal analysis in differentiating between subject groups.

Conclusions:

• The developed end-to-end dynamic spatiotemporal graph neural network model effectively utilizes both temporal and spatial information from rs-fMRI.
• This approach significantly improves classification performance for Alzheimer's disease, mild cognitive impairment, and normal cognition.
• The findings underscore the importance of integrating dynamic spatio-temporal analysis for accurate neuroimaging-based disease identification.