A realistic multimodal modeling approach for the evaluation of distributed source analysis: application to sLORETA.

OBJECTIVE: Electrical source localization (ESL) deriving from scalp EEG and, in recent years, from intracranial EEG (iEEG), is an established method in epilepsy surgery workup. We aimed to validate the distributed ESL derived from scalp EEG and iEEG, particularly regarding the spatial extent of the source, using a realistic epileptic spike activity simulator. APPROACH: ESL was applied to the averaged scalp EEG and iEEG spikes of two patients with drug-resistant structural epilepsy. The ESL results for both patients were used to outline the location and extent of epileptic cortical patches, which served as the basis for designing a spatiotemporal source model. EEG signals for both modalities were then generated for different anatomic locations and spatial extents. ESL was subsequently performed on simulated signals with sLORETA, a commonly used distributed algorithm. ESL accuracy was quantitatively assessed for iEEG and scalp EEG. MAIN RESULTS: The source volume was overestimated by sLORETA at both EEG scales, with the error increasing with source size, particularly for iEEG. For larger sources, ESL accuracy drastically decreased, and reconstruction volumes shifted to the center of the head for iEEG, while remaining stable for scalp EEG. Overall, the mislocalization of the reconstructed source was more pronounced for iEEG. SIGNIFICANCE: We present a novel multiscale framework for the evaluation of distributed ESL, based on realistic multiscale EEG simulations. Our findings support that reconstruction results for scalp EEG are often more accurate than for iEEG, owing to the superior 3D coverage of the head. Particularly the iEEG-derived reconstruction results for larger, widespread generators should be treated with caution.

Brain (Hyper)Excitability Revealed by Optimal Electrical Stimulation of GABAergic Interneurons.

BACKGROUND: Neurological disorders are often characterized by an excessive and prolonged imbalance between neural excitatory and inhibitory processes. An ubiquitous finding among these disorders is the disrupted function of inhibitory GABAergic interneurons. OBJECTIVE: The objective is to propose a novel stimulation procedure able to evaluate the efficacy of inhibition imposed by GABAergic interneurons onto pyramidal cells from evoked responses observed in local field potentials (LFPs). METHODS: Using a computational modeling approach combined with in vivo and in vitro electrophysiological recordings, we analyzed the impact of electrical extracellular, local, bipolar stimulation (ELBS) on brain tissue. We implemented the ELBS effects in a neuronal population model in which we can tune the excitation-inhibition ratio and we investigated stimulation-related parameters. Computer simulations led to sharp predictions regarding: i) the shape of evoked responses as observed in local field potentials, ii) the type of cells (pyramidal neurons and interneurons) contributing to these field responses and iii) the optimal tuning of stimulation parameters (intensity and frequency) to evoke meaningful responses. These predictions were tested in vivo (mouse). Neurobiological mechanisms were assessed in vitro (hippocampal slices). RESULTS: Appropriately-tuned ELBS allows for preferential activation of GABAergic interneurons. A quantitative neural network excitability index (NNEI) is proposed. It is computed from stimulation-induced responses as reflected in local field potentials. NNEI was used in four patients with focal epilepsy. Results show that it can readily reveal hyperexcitable brain regions. CONCLUSION: Well-tuned ELBS and NNEI can be used to locally probe brain regions and quantify the (hyper)excitability of the underlying brain tissue.

Recording of fast activity at the onset of partial seizures: depth EEG vs. scalp EEG.

Rapid discharges (25-80 Hz), a characteristic EEG pattern often recorded at seizure onset in partial epilepsies, are often considered as electrophysiological signatures of the epileptogenic zone. While the recording of rapid discharges from intracranial electrodes has long been established, their observation from the scalp is challenging. The prevailing view is that rapid discharges cannot be seen clearly (or at all) in scalp EEG because they have low signal-to-noise ratio. To date, however, no studies have investigated the 'observability' of rapid discharges, i.e. under what conditions and to what extent they can be visible in recorded EEG signals. Here, we used a model-based approach to examine the impact of several factors (distance to sources, skull conductivity, source area, source synchrony, and background activity) on the observability of rapid discharges in simultaneously simulated depth EEG and scalp EEG signals. In our simulations, the rapid discharge was clearly present in depth EEG signals but mostly almost not visible in scalp EEG signals. We identified some of the factors that may limit the observability of the rapid discharge on the scalp. Notably, surrounding background activity was found to be the most critical factor. The findings are discussed in relation to the presurgical evaluation of epilepsy.

The neuronal sources of EEG: modeling of simultaneous scalp and intracerebral recordings in epilepsy.

In many applications which make use of EEG to investigate brain functions, a central question is often to relate the recorded signals to the spatio-temporal organization of the underlying neuronal sources of activity. A modeling attempt to quantitatively investigate this imperfectly known relationship is reported. The proposed plausible model of EEG generation relies on an accurate representation of the neuronal sources of activity. It combines both an anatomically realistic description of the spatial features of the sources (convoluted dipole layer) and a physiologically relevant description of their temporal activities (coupled neuronal populations). The model was used in the particular context of epileptiform activity (interictal spikes) to interpret simultaneously generated scalp and intracerebral EEG. Its integrative properties allowed for the bridging between source-related parameters (spatial extent, location, synchronization) and the properties of resulting EEG signals (amplitude of spikes, amplitude gradient along intracerebral electrodes, topography over scalp electrodes). The sensitivity of both recording modalities to source-related parameters was also studied. The model confirmed that the cortical area involved in interictal spikes is rather large. Its relative location with respect to recording electrodes was found to strongly influence the properties of EEG signals as the source geometry is a critical parameter. The influence, on simulated signals, of the synchronization degree between neuronal populations within the epileptic source was also investigated. The model revealed that intracerebral EEG can reflect epileptic activities corresponding to weak synchronization between neuronal populations of the epileptic patch. These results, as well as the limitations of the model, are discussed.

Modeling and interpretation of scalp-EEG and depth-EEG signals during interictal activity.

In epileptic patients candidate to surgery, the interpretation of electrophysiological signals recorded invasively (depth-EEG) and non-invasively (scalp-EEG) is a crucial issue to determine epileptogenic network and to define subsequent therapeutic strategy. This issue is addressed in this work through realistic modeling of both scalp-EEG and depth-EEG signals. The model allows for studying the influence, on signals, of source-related parameters leading to the generation of epileptic transient activity (interictal spikes). This parametric study is based on a variety of scenarios in which either spatial or temporal features of the sources of activity are modified. Statistical quantities measured on simulated signals allow for better understanding of the influence of source-related parameters on the information conveyed by these signals, collected from scalp or depth electrodes.

A realistic spatiotemporal source model for EEG activity: application to the reconstruction of epileptic depth-EEG signals.

The context of this work is the interpretation of depth-EEG signals recorded in epileptic patients. This study focuses on the relationship between spatial and temporal properties of neuronal sources and depth-EEG signals observed along intracerebral electrodes (source/sensor relationship). We developed an extended source model which connects two levels of representation: a model of coupled neuronal populations and a distributed current dipole model. This model was used to simulate epileptic spiking depth-EEG signals from the forward solution at each intracerebral sensor location. Results showed that realistic spikes were obtained in the model under two specific conditions: a sufficiently large spatial extension of the neocortical source and a high degree of coupling between activated neuronal populations composing this extended source.