Classification of Mild Cognitive Impairment Based on Dynamic Functional Connectivity Using Spatio-Temporal...

Area of Science:

• Neuroimaging
• Artificial Intelligence
• Computational Neuroscience

Background:

• Dynamic functional connectivity (dFC) captures neural activity changes via resting-state functional MRI (rs-fMRI).
• Existing dFC analyses often underutilize sequential information crucial for identifying brain conditions like Alzheimer's disease (AD).

Purpose of the Study:

• To propose a novel framework for joint learning of spatial and temporal information within dFC using a transformer architecture.
• To enhance the early prediction of Mild Cognitive Impairment (MCI), a precursor to AD.

Main Methods:

• rs-fMRI data were used to construct dFC networks via a sliding window approach.
• A transformer-based framework with temporal and spatial blocks was employed to capture dynamic spatio-temporal dependencies.
• Contrastive learning was integrated to improve feature representation robustness and reduce reliance on labeled data.

Main Results:

• The proposed method demonstrated superior performance in predicting MCI.
• Experimental results were validated on data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort.
• The framework effectively leverages sequential dFC information for improved diagnostic accuracy.

Conclusions:

• The novel transformer-based framework offers a powerful approach for analyzing dFC data.
• This method shows significant potential for the early identification and prediction of Alzheimer's disease progression.
• Integrating spatial and temporal analysis with contrastive learning enhances the utility of rs-fMRI in neurodegenerative disease research.