Clinical Prediction of Functional Decline in Multiple Sclerosis Using Volumetry-Based Synthetic Brain Networks.

Area of Science:

• Neuroscience
• Medical Imaging
• Machine Learning

Background:

• Multiple sclerosis (MS) disability progression is linked to large-scale brain network disruption.
• Diffusion tensor imaging (DTI) is limited in clinical use for assessing structural connectivity.
• Alternative MRI methods are needed to capture network alterations.

Purpose of the Study:

• To determine if synthetic structural connectivity from volumetric MRI can detect MS network alterations.
• To assess if this method predicts functional decline, especially in upper limbs.
• To provide a clinically accessible tool for MS connectomic analysis.

Main Methods:

• Generated synthetic structural connectivity matrices from routine T1-weighted MRI.
• Integrated structural covariance, distance-dependent connectivity, and disease patterns.
• Applied graph theory and machine learning to predict clinical progression (9-Hole Peg Test).

Main Results:

• Synthetic networks showed plausible organization; global efficiency correlated with disability.
• Clinical progression linked to reduced network integration/segregation and increased path length.
• Machine learning models accurately predicted upper limb decline (balanced accuracy >80%, AUC-ROC up to 0.85).

Conclusions:

• Synthetic structural connectivity from volumetric MRI effectively captures MS network disruption.
• This approach enables accurate prediction of functional decline in MS patients.
• Provides a feasible alternative to DTI for network analysis in MS.