Assessment of Effective Network Connectivity among MEG None Contaminated Epileptic Transitory Events

Area of Science:

• Neuroscience
• Biomedical Engineering
• Medical Imaging

Background:

• Epileptogenic zones (EZ) are crucial for epilepsy diagnosis.
• Magnetoencephalography (MEG) and inverse problem techniques are used to locate abnormal brain activity.
• Different inverse problem methods have varying assumptions and network connectivity estimations.

Purpose of the Study:

• To evaluate the performance of distributed inverse problem techniques in defining EZ.
• To compare the network connectivity derived from MEG with intracranial EEG (iEEG).
• To assess the clinical utility of these techniques in identifying seizure onset zones.

Main Methods:

• Applied Singular Value Decomposition (SVD) to isolate interictal epileptiform discharges from MEG data.
• Validated the SVD technique on simulated and patient MEG data.
• Utilized four inverse problem methods (cMEM, eLORETA, dSPM, MNE) to identify cortical sources of excessive discharges.
• Computed and compared network connectivity from MEG inverse solutions with iEEG data.

Main Results:

• Coherent Maximum Entropy on the Mean (cMEM) demonstrated the highest agreement with iEEG network connectivity, particularly in source distance.
• Exact low-resolution brain electromagnetic tomography (eLORETA) and Dynamical Statistical Parametric Mapping (dSPM) showed the strongest connection strengths.
• Clinical performance showed an average detection of 73.5% of deep sources in MEG and 77.15% of MEG sources in iEEG.
• eLORETA provided the lowest propagation delay (18 ms), closely matching iEEG.

Conclusions:

• Distributed inverse problem techniques can effectively identify parts of the seizure onset zone.
• cMEM and eLORETA are promising methods for accurate EZ localization and network analysis in epilepsy.
• These findings can assist neurologists in improving epilepsy diagnosis and treatment planning.