User-defined virtual sensors: A new solution to the problem of temporal plus epilepsy sources

Area of Science:

• Neuroscience
• Medical Imaging
• Epileptology

Background:

• Medically resistant epilepsy (MRE) often involves the temporal lobe (TLE).
• Temporal plus epilepsy (TLE+) patients face a five-fold higher risk of surgical failure.
• Accurate pre-surgical assessment is crucial for optimizing epilepsy surgery outcomes.

Purpose of the Study:

• To correlate magnetoencephalography (MEG) virtual sensor waveform analyses with surgical outcomes in MRE patients.
• To compare visual and computational MEG analyses with invasive EEG (iEEG) findings.
• To evaluate the utility of user-defined virtual sensor (UDvs) beamforming and effective connectivity (EC) hubs in classifying epilepsy types (TLE vs. TLE+) and predicting surgical success.

Main Methods:

• Retrospective analysis of 80 MRE patients who underwent MEG and iEEG monitoring.
• Application of UDvs beamforming with sensors placed in key brain regions.
• Computation of MEG effective connectivity (EC) using eigenvector centrality to identify hub regions.
• Comparison of UDvs beamformer and EC hubs with iEEG data and surgical outcomes.

Main Results:

• UDvs beamforming showed a significant association with both postsurgical outcome (OR=1.22) and epilepsy classification (TLE vs. TLE+) (OR=1.47).
• EC hub location was significantly associated with a good postsurgical outcome (OR=1.22).
• Conventional MEG methods did not yield significant results compared to UDvs beamformer.

Conclusions:

• UDvs beamforming demonstrates concordance with iEEG in predicting postsurgical seizure outcome and classifying epilepsy types.
• UDvs beamforming offers a complementary approach to established methods like equivalent current dipole (ECD) analysis.
• This technique can potentially enhance invasive electrode and surgical resection planning for epilepsy surgery evaluations.