Improving Detection of Hippocampal Epileptiform Activity Using Magnetoencephalography

Purpose: Magnetoencephalography (MEG) defines the spike-generating zone and provides targets for invasive monitoring with stereotactic electroencephalography. This retrospective, blinded, cross-sectional study determined whether MEG virtual sensors could identify hippocampal epileptiform activity. Methods: Using MEG beamformer analysis, virtual sensors were manually placed in bilateral hippocampi and corresponding virtual sensor waveforms were analyzed for the presence of epileptiform activity. These findings were compared with hippocampal stereotactic electroencephalography in the same patients. Concordance was determined using sensitivity and specificity. Results: Thirty patients (mean age 12.5 +/- 5.9 years) and 35 hippocampi were included. Patients were also placed into subgroups based on conventional MEG analysis: temporal (n = 19), extratemporal (n = 10), and normal (n = 1). Overall, sensitivity and specificity were 57.9% and 50.0%, respectively (n = 35). Patients with temporal sources based on conventional MEG analysis had sensitivity and specificity of 80.0% and 36.4%, respectively (n = 21). Those with extratemporal sources based on conventional MEG had sensitivity and specificity of 42.9% and 80.0%, respectively (n = 12). Conclusions: When grouped by conventional MEG analysis, virtual sensors can be useful to confirm mesial temporal dipoles seen with conventional analysis. Significance: This work may help support the use of MEG for the detection of epileptiform activity in the hippocampus and influence the planning of invasive electrode placement.