Identification of epileptogenic networks from dense EEG: A model-based study

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Epilepsy is recognized as a network disorder, making the identification of epileptogenic networks challenging.
• Magnetoencephalography (M/EEG) source connectivity offers high temporal and spatial resolution for mapping brain networks.
• Noninvasive recordings are essential for understanding epilepsy network dynamics.

Purpose of the Study:

• To analyze the impact of EEG inverse problem solutions and functional connectivity measures on identifying epileptogenic networks.
• To compare the performance of different inverse algorithms (dSPM, wMNE, cMEM) and connectivity metrics (r(2), h(2), MI).

Main Methods:

• Simulated realistic interictal epileptic spikes using a combined biophysical/physiological model.
• Evaluated three inverse solutions: dSPM, wMNE, and cMEM.
• Assessed three functional connectivity measures: r(2) (linear), h(2), and MI (nonlinear).

Main Results:

• The combination of inverse solution and connectivity measure significantly impacts network identification.
• Nonlinear connectivity measures (h(2), MI) demonstrated higher efficiency than linear measures (r(2)).
• The dSPM inverse solution exhibited lower performance compared to cMEM and wMNE.

Conclusions:

• The selection of EEG inverse solutions and connectivity measures critically influences the accuracy of epileptogenic network identification.
• Nonlinear connectivity analysis is more effective for detecting epileptic networks from M/EEG data.
• wMNE and cMEM are preferable inverse solutions over dSPM for this application.