Objective:

This study observes whether user-defined virtual sensors (UDvs) beamforming, using expert reader analysis, is superior to the conventional equivalent current dipole (ECD) for non-invasively differentiating between TLE and TLE+. This could result in decreased failure rates in temporal lobe epilepsy surgery and improved seizure-free outcomes. 

Background:

Temporal lobe epilepsy (TLE) is the most common seizure disorder. 44% of TLE+ pediatric patients that undergo resective surgery fail to achieve seizure freedom. It is imperative to identify TLE vs TLE+ in children pre-surgically, since the invasive monitoring plan to identify the seizure onset zone (SOZ) would differ. 

Design/Methods:

Ten patients who underwent magnetoencephalography (MEG) as part of non-invasive pre-surgical evaluation and intracranial electroencephalography (iEEG) monitoring were included in this retrospective analysis. Five patients each were assigned to two groups: TLE and TLE+. Virtual sensors were placed symmetrically in the bilateral mesial temporal structures, temporal gyri, and neighboring structures. We reconstructed virtual sensor waveforms and analyzed three ten-minute segments of MEG recordings to classify them as positive or negative for epileptiform activity. A chart review of iEEG data was also classified as either positive or negative spikes. The primary analysis calculated the sensitivity and specificity of each method to determine the predictive value of UDvs-beamforming.

Results:

Ten patients (mean age 9.8 ± 4.2 years) with TLE or TLE+ were included. Across the ten patients, the UDvs method had a mean sensitivity of 93.1% and specificity of 53.3% compared to the ECD method which had a sensitivity of 71.1% and specificity of 88.3% (P = 0.031, 0.002). 

Conclusions:

The ECD method has greater specificity than UDvs while UDvs had greater sensitivity than ECD in determining the presence of interictal epileptic spikes compared to the gold standard, iEEG. Future work will examine increasing sample size, performing connectivity analysis of interictal spikes, and correlating surgical resection with patient outcomes.