Dynamic Network Connectivity Analysis to Identify Epileptogenic Zones Based on Stereo-Electroencephalography

Area of Science:

• Neuroscience
• Epileptology
• Network Science

Background:

• Accurate localization of epileptogenic zones (EZs) is critical for successful epilepsy surgery.
• Refractory focal epilepsy poses significant challenges for treatment.
• Stereo-electroencephalography (SEEG) provides detailed neural activity data.

Purpose of the Study:

• To evaluate the effectiveness of dynamic network connectivity analysis using SEEG signals for localizing EZs.
• To determine if graph theory measures can identify critical nodes within the epileptic brain network.

Main Methods:

• SEEG data from seven epilepsy patients with successful surgical outcomes were analyzed.
• A time-variant multivariate autoregressive model with a Kalman filter was employed.
• Dynamic directed network models were constructed using partial directed coherence.
• In-degree, out-degree, and betweenness centrality were calculated to analyze network characteristics.

Main Results:

• In-degree and betweenness centrality effectively localized EZs in all seven patients, correlating well with clinical diagnoses.
• Out-degree analysis did not reveal significant differences between network nodes.
• The dynamic network connectivity approach demonstrated high accuracy in EZ identification.

Conclusions:

• Ictal SEEG signals combined with effective connectivity analysis provide an accurate method for EZ localization.
• In-degree and betweenness centrality are superior network characteristics for identifying EZs compared to out-degree.
• This approach offers a promising tool for presurgical evaluation in refractory focal epilepsy.