Characterization of relapsing-remitting multiple sclerosis patients using support vector machine classifications of...

Area of Science:

• Neuroimaging
• Machine Learning
• Biomarker Discovery

Background:

• Clinical symptoms of Multiple Sclerosis (MS) poorly correlate with traditional MRI findings.
• Diagnosis and progression monitoring of MS rely on combined MRI and clinical assessment.
• Novel biomarkers are crucial for a comprehensive understanding of MS.

Purpose of the Study:

• To employ machine learning for classifying relapsing-remitting MS patients and healthy controls.
• To identify significant brain regions and connectivity measures for MS characterization.
• To assess the utility of structural and functional connectivity in differentiating MS patient disability levels.

Main Methods:

• Acquisition of MRI data from MS patients and healthy subjects.
• Extraction of fractional anisotropy maps, structural, and functional connectivity.
• Construction of support vector machine classifiers using individual and combined connectivity features.
• Comparison of three group pairs: MS patients vs. controls, and two patient disability groups.

Main Results:

• Classifiers achieved up to 89% accuracy in distinguishing MS patients from healthy subjects.
• Classification performance was significantly lower (<63%) when differentiating between MS patient disability groups.
• Key brain regions identified include the right occipital, left frontal orbital, medial frontal cortices, and lingual gyrus.

Conclusions:

• Machine learning models based on MRI data reliably differentiate MS patients from healthy individuals.
• The study successfully identified critical brain regions and connectivity patterns relevant to MS.
• Further research is needed to refine models for assessing MS patient disability progression.