Data-Driven Topological Filtering Based on Orthogonal Minimal Spanning Trees: Application to Multigroup...

Area of Science:

• Neuroscience
• Network Science
• Computational Biology

Background:

• Functional brain networks are crucial for understanding brain function and dysfunction.
• Existing methods for network analysis may not optimally capture essential connections for information flow.

Purpose of the Study:

• To introduce a novel data-driven topological filtering technique, orthogonal minimal spanning trees (OMSTs), for deriving functional brain network backbones.
• To identify essential functional connections that ensure optimal information flow using a criterion of global efficiency minus connection cost.
• To evaluate the efficacy of the OMST technique in discriminating between different participant groups.

Main Methods:

• Application of the OMST technique to multichannel, resting-state magnetoencephalography (MEG) recordings.
• Computation of weighted interactions between network nodes using dominant intrinsic coupling modes with symbolic mutual information and phase lag index.
• Comparison of OMST-derived network classification performance against relative power spectrum features and conventional minimal spanning tree algorithms.

Main Results:

• The OMST technique demonstrated excellent discrimination of individual cases based on group membership.
• Classification accuracy using OMST-derived functional networks significantly outperformed conventional methods.
• The method successfully identified essential functional connections for optimal information flow.

Conclusions:

• The OMST technique provides a robust and effective method for analyzing functional brain networks.
• This approach offers superior performance in classifying individuals across different neurological and developmental conditions.
• OMSTs represent a promising tool for advancing the study of brain connectivity and its clinical applications.