Epileptic fast intracerebral EEG activity: evidence for spatial decorrelation at seizure onset

Abstract

Low-voltage rapid discharges (or fast EEG ictal activity) constitute a characteristic electrophysiological pattern in focal seizures of human epilepsy. They are characterized by a decrease of signal voltage with a marked increase of signal frequency (typically beyond 25 Hz). They have long been observed in stereoelectroencephalographic (SEEG) signals recorded with intra-cerebral electrodes, generally occurring at seizure onset and simultaneously involving distinct brain regions. Spectral properties of rapid ictal discharges as well as spatial correlations measured between SEEG signals generated from distant sites before, during and after these discharges were studied. Cross-correlation estimates within typical EEG sub-bands and statistical tests performed in 10 patients suffering from partial epilepsy (frontal, temporal or fronto-temporal) reveal that SEEG signals are significantly de-correlated during the discharge period compared with periods that precede and follow this discharge. These results can be interpreted as a functional decoupling of distant brain sites at seizure onset followed by an abnormally high re-coupling when the seizure develops. They lead to the concept of 'disruption' that is complementary of that of 'activation' (revealed by significantly high correlations between signals recorded during seizures), both giving insights into our understanding of pathophysiological processes involved in human partial epilepsies as well as in the interpretation of clinical semiology.

Figure 1

(a) example of intracerebral implantation scheme defined for SEEG exploration in patient POM from non invasive data (EEG, semiology, medical imaging), (b) Intracerebral multiple lead electrodes are implanted under stereotactic conditions as shown by the solid line rectangle that corresponds to Talairach’s reference frame. In our group, electrodes are identified by one or two capital letters (A, PO, …) and recording leads are numbered from 1 to 10 or 15 depending on the number of leads per electrode. Low weight numbers (1, 2, 3, …) generally correspond to deepest structures (for instance, leads A1, A2 record signals from the amygdala while A9, A10 record the lateral neocortex of the middle temporal gyrus). Bipolar signals are obtained from subtraction of signals recorded on two adjacent leads (for instance, bipolar signal A1-A2 may be obtained from monopolar signals A1 and A2).

Figure 2

a) example of intracerebral recording performed during SEEG exploration (patient TAL). Seizure begins precisely during the abrupt change in the activity reflected by signals (transition from sustained discharge of high amplitude spikes to a low voltage rapid discharge expressing in the gamma band of the EEG, as revealed by power spectral densities computed before and after seizure onset), b) Γ-activity map computed over a seven second period starting from seizure onset. A color-coding is used to represent the power in the upper frequency band (24 to 128 Hz) as a function of both time and space (recording channels). In this example, on three electrodes (AC, TR and OP) respectively recording the superior frontal sulcus (BA9 and BA9/46) and operculo-frontal region, bipolar signals AC7-AC8, TR7-TR8 and OP7-OP8 exhibit fast activity revealed by high power in the Γ sub-band (red color). A depicted on the map, very fast oscillations may appear quasi-simultaneously in distinct neural sites.

Figure 3

Signal processing procedure used to characterize the power of signals reflecting low voltage rapid discharge at seizure onset and used to quantify their spatio-temporal correlation. Single signals (a) are first studied for spectral features. A filter bank (Hamming filters, see text §2.2.1) is used to separate theta, alpha, beta and gamma (24 – 128 Hz) components in corresponding sub-bands (see table 2). Three periods (BD: before the discharge, D: during the discharge, AD: after the discharge) are defined by visual analysis. On each period, the power in each sub-band is estimated and represented as a color-coded map (b) aligned with the signal. Power in each sub-band is then averaged over each period (BD, D and AD). An histogram (e) representing the relative power per sub-band as a function of the period is then constructed. Spatio-temporal correlations are measured on each pair of signals. They are estimated via the computation of a linear correlation coefficient optimized for time delays (referred to as the r² coefficient). This computation is performed over two windows sliding on the two signals under analysis and in each sub-band and. A color-coded map (c) that gives r² values as a function of time and frequency sub-band is then built and aligned with signals. r² values range in the interval [0, 1], as shown in the plot (d) displaying values measured over the theta sub-band. Finally, r² values are averaged over the three periods of interest (BD, D and AD) in each frequency band and summarized by an histogram (f).

Figure 4

Relative power and spatio-temporal correlations measured in the ten studied patients. In all patients, power distribution changes as a function of the considered period (upper frequency band Γ becomes predominant during the low voltage rapid discharge). In all patients, r² values estimated over the discharge period are lower than those measured before and after this period, denoting the appearance of a spatial decorrelation.

Figure 5

Results of statistical analysis performed to test the significance of the observed decrease of correlation values measured during the discharge period. a) r² values summed over the frequency bands are first normalized in order to make their distribution as Gaussian (b) before performing statistical tests. c) Boxplot and d) t-test performed on normalized values demonstrate that r² values measured during the discharge are significantly lower than those measured before and after and values measured after the discharge are significantly higher than those measured before discharge (see text for details about the interpretation of boxplot features in the studied context).