Noninvasive Mapping of Ripple Onset Predicts Outcome in Epilepsy Surgery

Abstract

Objective: Intracranial electroencephalographic (icEEG) studies show that interictal ripples propagate across the brain of children with medically refractory epilepsy (MRE), and the onset of this propagation (ripple onset zone [ROZ]) estimates the epileptogenic zone. It is still unknown whether we can map this propagation noninvasively. The goal of this study is to map ripples (ripple zone [RZ]) and their propagation onset (ROZ) using high-density EEG (HD-EEG) and magnetoencephalography (MEG), and to estimate their prognostic value in pediatric epilepsy surgery. Methods: We retrospectively analyzed simultaneous HD-EEG and MEG data from 28 children with MRE who underwent icEEG and epilepsy surgery. Using electric and magnetic source imaging, we estimated virtual sensors (VSs) at brain locations that matched the icEEG implantation. We detected ripples on VSs, defined the virtual RZ and virtual ROZ, and estimated their distance from icEEG. We assessed the predictive value of resecting virtual RZ and virtual ROZ for postsurgical outcome. Interictal spike localization on HD-EEG and MEG was also performed and compared with ripples. Results: We mapped ripple propagation in all patients with HD-EEG and in 27 (96%) patients with MEG. The distance from icEEG did not differ between HD-EEG and MEG when mapping the RZ (26-27mm, p = 0.6) or ROZ (22-24mm, p = 0.4). Resecting the virtual ROZ, but not virtual RZ or the sources of spikes, was associated with good outcome for HD-EEG (p = 0.016) and MEG (p = 0.047). Interpretation: HD-EEG and MEG can map interictal ripples and their propagation onset (virtual ROZ). Noninvasively mapping the ripple onset may augment epilepsy surgery planning and improve surgical outcome of children with MRE.

Case Report: Laser Ablation Guided by State of the Art Source Imaging Ends an Adolescent's 16-Year Quest for Seizure Freedom

“…Among others, our group has shown that HFOs are promising interictal biomarkers of epilepsy, which can be recorded non-invasively with scalp EEG and MEG (Papadelis et al, 2009(Papadelis et al, , 2016Tamilia et al, 2017;Zijlmans et al, 2017;Thomschewski et al, 2019;Thomschewski et al, 2021). Surgical resection of the non-invasively identified HFO-zone predicts surgical outcome (Tamilia et al, 2021). Here, we localized an incident of HFO-activity observed in both conv-EEG and MEG and localized with MSI.…”

“…In a recent review paper, Zijlmans et al (2019) proposed the use of electric and magnetic source imaging (ESI/MSI) obtained from high-density EEG (HD-EEG) or magnetoencephalography (MEG) for almost all electroclinical syndromes. Their proposal was based on findings from several research groups, including ours, showing that ESI and MSI can provide valuable clinical information in the presurgical evaluation of patients with DRE, evaluated for surgical therapy (Mégevand et al, 2014;Papadelis et al, 2016;Tamilia et al, 2017Tamilia et al, , 2019Tamilia et al, , 2020Tamilia et al, , 2021Plummer et al, 2019;Rampp et al, 2019;Ntolkeras et al, 2021;Papadelis and Perry, 2021;Ricci et al, 2021). ESI and MSI can provide a precise estimate of the irritative zone that can be interpreted as presurgical indicator of the EZ location (Tamilia et al, 2019).…”

Changes in the Functional Brain Network of Children Undergoing Repeated Epilepsy Surgery: An EEG Source Connectivity Study

“…We must also mention that non-invasive recording techniques with a higher number of sensors [ 56 ], such as high-density EEG (>64 channels) or MEG [ 47 , 57 ], allow a more precise reconstruction of the network through smaller ROIs (also called virtual sensors). Most clinics still lack the equipment or resources to perform MEG or EEG monitoring with >64 electrodes: thus, while our proposed approach can be significantly enhanced in terms of localizing value, it paves the way for an easily implementable approach in clinics.…”

“…Source space was constrained to the cortex, which was composed of ~15,000 vertices per patient. We used a linearly constrained minimum variance (LCMV) beamformer to estimate brain activity within the source space [ 47 ]. Data covariance was computed from the EEG epochs, whereas identity matrix was used for noise covariance.…”

Graph Theoretical Measures of Fast Ripples Support the Epileptic Network Hypothesis.

“…Recently, it has also been suggested that HFOs sites are functionally connected which may be important in predicting response to epilepsy surgery [33][34][35] .…”