Multilayer Network Approach in EEG Motor Imagery with an Adaptive Threshold.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Network Science

Background:

• The brain is commonly modeled as a neural network, often using single-layer graphs to represent its topology and dynamics.
• Single-layer models have limitations in capturing the full complexity of brain processes.
• Multilayer modeling offers a more comprehensive approach by integrating diverse features for whole-brain analysis.

Purpose of the Study:

• To analyze electroencephalography (EEG) dynamics during motor imagery tasks using both single-layer and multilayer network modeling.
• To compare the effectiveness of traditional fixed thresholds versus a novel adaptive threshold (Otsu's method) for constructing reliable network topologies.

Main Methods:

• Utilized a database of 18 EEG recordings from four motor imagery tasks (left hand, right hand, feet, tongue).
• Estimated brain connectivity by calculating coherence adjacency matrices across different electrophysiological bands (δ, θ, α, β).
• Modeled each frequency band as a single-layer graph and as a layer within a multilayer brain network model, applying both fixed and Otsu's adaptive thresholds.

Main Results:

• Multilayer network modeling provides a more integrated approach to whole-brain analysis compared to single-layer models.
• Otsu's adaptive thresholding method offers a reliable way to distinguish effective neural connections from spurious ones in brain network construction.
• Analysis of the brain network models indicated significant involvement of frontal and parietal brain regions during motor imagery tasks.

Conclusions:

• Multilayer network modeling is a powerful tool for dissecting complex brain dynamics, particularly in the context of motor imagery.
• The proposed adaptive thresholding method enhances the reliability of brain network topology construction.
• Frontal and parietal regions are crucial for the neural processing underlying motor imagery.