Authors :
Presenting Author: Ilana Lefkovitz, MD – Cleveland Clinic
Ahsan Moosa Naduvil Valappil, MD, FAES – Cleveland Clinic
William Bingaman, MD – Cleveland Clinic
Balu Krishnan, PhD – Cleveland Clinic
Hiroatsu Murakami, MD, PhD – Cleveland Clinic
Rationale:
Magnetoencephalography (MEG) is a non-invasive modality that records the brain’s magnetic fields to identify epileptiform activity. While source modeling of epileptiform discharges is routinely used to localize the irritative zone, functional connectivity analysis offers a complementary approach to study information flow during interictal and ictal phases. We present two pediatric cases in which functional connectivity analysis provided insights beyond MEG dipole findings.
Methods:
MEG data was acquired using a 306 channel whole-head Elekta system and processed with the temporally extended signal space separation algorithm (tSSS). Interictal spikes were localized using the single equivalent current dipole method and dipole locations were clustered using the k-means algorithm. MEG segments were extracted around interictal spikes, non-epileptiform background, and ictal activity. Source modeling and connectivity analysis was performed. Regions of interest (ROIs) were manually drawn on the cortical surface and average source time series were computed. Directional connectivity was assessed using the Directed Transfer Function (DTF). Spike segments were duration-matched to background and ictal connectivity matrices were normalized to their corresponding background segments.
Results:
The first patient, a girl with Rasmussen’s encephalitis and drug-resistant epilepsy, had bilateral but right-predominant spike-wave discharges on EEG and right perisylvian and perirolandic atrophy on MRI. She underwent a right functional hemispherectomy, but seizures recurred. Postoperatively, MEG dipoles indicated interictal and ictal activity from the left posterior perisylvian and cingulate regions, raising concern for left-sided epileptogenicity. She then underwent a right anatomic hemispherectomy and is now seizure-free. Retrospective connectivity analysis showed increased interictal outflow from the right medial orbitofrontal and mid-cingulate cortex. In the ictal phase, outflow shifted to the right anterior frontal region preictally, then to the anterior cingulate during the seizure, without spread to the left hemisphere—thus matching the resected area.
The second patient, a boy with a right MCA perinatal stroke and drug-resistant epilepsy, initially showed right centroparietal discharges on EEG and right frontal MEG dipoles, leading to a right frontal disconnection. Seizures worsened. Repeat MEG revealed interictal dipoles in the left inferior frontal and frontal opercular regions and ictal dipoles in the left precentral sulcus/gyrus. One interictal dipole was noted in the right frontal region. Connectivity analysis showed increased interictal outflow from the right medial frontal region, shifting to the left medial frontal region during seizures.
Conclusions:
These cases highlight the added value of MEG-based functional connectivity in localizing epileptogenic networks. Directional connectivity clarified ambiguous dipole findings and revealed network-level dynamics consistent with clinical outcome in the first case, while informing management in the second. Integrating connectivity metrics with conventional source modeling may improve presurgical evaluation in complex pediatric epilepsy.
Funding: None