Classification of Migraine Using Static Functional Connectivity Strength and Dynamic Functional Connectome Patterns:...

Area of Science:

• Neuroimaging
• Neurology
• Medical Diagnostics

Background:

• Migraine is a prevalent, chronic neurological disorder with poorly understood causes and a lack of objective diagnostic markers.
• Resting-state functional magnetic resonance imaging (RS-fMRI) is a promising tool for brain disorder research, but high-dimensional data presents challenges.
• Accurate classification and diagnosis of migraine remain difficult due to these limitations.

Purpose of the Study:

• To develop and evaluate an automatic recognition framework for classifying migraine sufferers from healthy controls.
• To compare the diagnostic utility of static functional connectivity (sFC) strength features versus dynamic functional connectome pattern (DFCP) features.
• To assess the combined performance of sFC and DFCP features for enhanced migraine classification.

Main Methods:

• Extracted static functional connectivity (sFC) strength and dynamic functional connectome pattern (DFCP) features from RS-fMRI data.
• Employed recursive feature elimination with support vector machine (SVM-RFE) for optimal feature selection.
• Utilized a support vector machine (SVM) classifier for training and testing the diagnostic framework.

Main Results:

• Dynamic functional connectome pattern (DFCP) features demonstrated significantly higher classification performance compared to static functional connectivity (sFC) strength features.
• The combination of sFC strength and DFCP features yielded the optimal classification performance, highlighting synergistic benefits.
• The proposed framework showed robust performance in differentiating individuals with migraine from healthy controls.

Conclusions:

• Dynamic functional connectome patterns offer a significant advantage for migraine classification over static functional connectivity.
• Combining static and dynamic functional connectivity features optimizes diagnostic performance in migraine.
• The developed RS-fMRI based classification framework shows potential for diagnosing migraine and possibly other neurological disorders.