Alzheimer Classification Using a Minimum Spanning Tree of High-Order Functional Network on fMRI Dataset

Area of Science:

• Neuroimaging
• Network Neuroscience
• Computational Psychiatry

Background:

• Functional magnetic resonance imaging (fMRI) is crucial for mapping brain functional connectivity.
• Conventional methods often overlook dynamic changes in connectivity, impacting network analysis.
• Previous high-order network approaches faced limitations in classification accuracy and interpretability due to clustering methods.

Purpose of the Study:

• To develop a novel method for constructing high-order functional connectivity networks that preserves dynamic characteristics and neurological interpretability.
• To introduce the minimum spanning tree (MST) method for simplifying complex network structures.
• To enhance the diagnostic accuracy of resting-state fMRI for Alzheimer's disease.

Main Methods:

• Implementation of the minimum spanning tree (MST) method on high-order functional connectivity networks derived from resting-state fMRI data.
• Development of a multi-parameter optimization framework for extracting discriminative features from MST-based networks.
• Comparison of the proposed method against conventional network generation techniques.

Main Results:

• The MST method effectively simplifies high-order network structures while retaining core connectivity information.
• Dynamic characteristics of brain activity time series are preserved, ensuring neurological interpretability.
• The proposed resting-state fMRI classification method demonstrated significantly improved diagnostic accuracy for Alzheimer's disease.

Conclusions:

• The minimum spanning tree approach offers an unbiased and effective way to analyze dynamic functional connectivity networks.
• This novel method enhances the reliability and interpretability of brain network analysis in neuroimaging.
• The findings suggest a promising avenue for improving early diagnosis of neurodegenerative diseases like Alzheimer's using fMRI.