Abstracts

Rationale: High Frequency Oscillations (HFOs) are emerging as biomarkers of epileptogenicity. Small brain regions produce HFOs. Given the high resistivity of the skull and the assumption that a large extent of cortex is needed to observe an event on the scalp, HFOs have been mostly recorded with intracranial electrodes. Surprisingly, recent studies showed that spontaneous HFOs can be recorded from the scalp. How is it possible that these small extent events are visible on the scalp?What are the cortical correlates of scalp HFOs? Using simultaneous scalp and intracranial recordings, we studied the spatial distribution of scalp HFOs on the cortical surface. Based on simulations, we estimated the spatial extent on the scalp of focal cortical sources. Methods: Simultaneous scalp and subdural EEGs from 11 patients with focal epilepsy were processed (low-pass filtered at 344Hz and sampled at 1024Hz). Scalp HFOs and spikes were identified in 1 hour of night recording (10-20 system). Voltage maps at each event's maximum peak time were obtained and their similarity measured (cross-correlation, p<0.01). We applied a linear model to relate the observations on the subdural contacts to those on the scalp (leave one out procedure). To estimate the spatial distribution of a focal source on the scalp, we simulated (using Brainstorm) 275 distributed sources of 5nA.m situated over cortical gyri (area: 1.1cm2 +/-0.2cm2). Their spatial extent on the scalp was analyzed. Results: Scalp HFOs were observed in 9 of 11 patients (115 HFOs). Voltage maps on the subdural contacts were spatially extended at the time of scalp spikes, but were focal, consisting of one or a few dipolar configurations, at the time of scalp HFOs (Figure 1). Similar scalp maps corresponded to similar maps on the grids for 50% of the spikes, but only for 15% of the HFOs. These suggest that small cortical areas generated the HFOs seen on the scalp. The linear model between subdural and scalp contacts explained the scalp topography for 83% of the spikes but only for 38% of the HFOs. Thus, the assumption that a subdural grid "sees" everything that is being seen by the nearby scalp contacts is more valid for spikes than for HFOs. When simulating 275 cortical sources of approximately 1cm2, we found that we are spatially under-sampling these events with the 10-20 (17% visible above 5uV) or 10-10 (31%) systems (Figure 2). Conclusions: The spatial distribution on the cortical surface at the time of scalp HFOs was focal, with a spatial extent sometimes less than one square cm. Different subdural patterns were observed for similar scalp HFO patterns. Even though the generators of HFOs are small, they can be observed on the scalp, with low amplitude and in a very focal region. A dense distribution of scalp electrodes seems therefore necessary to fully spatially sample HFOs on the scalp. A better understanding of the spatial sampling needed to observe high frequency brain activity on the scalp is important for the clinical use of scalp HFOs as biomarkers of epilepsy as well as in cognitive research. Supported by Savoy Foundation, DAAD short-term scholarship and CIHR MOP-102710.