Brain Advance Access originally published online on April 8, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Brain, Vol. 126, No. 6, 1449-1459,
June 2003
© 2003 Guarantors of Brain
doi: 10.1093/brain/awg144
Epileptic fast intracerebral EEG activity: evidence for spatial decorrelation at seizure onset
1 Laboratoire Traitement du Signal et de L’Image, INSERM, Université de Rennes 1, Campus de Beaulieu, Rennes and 2 Laboratoire de Neurophysiologie et Neuropsychologie, INSERM, Université de la Méditerranée, Marseille, France
Correspondence to: F. Wendling, Laboratoire Traitement du Signal et de L’Image, Université de Rennes 1, INSERM, Campus de Beaulieu, Bat. 22, 35042 Rennes Cedex, France E-mail: fabrice.wendling{at}univ-rennes1.fr
Received November 15, 2002. Revised January 23, 2003. Accepted February 3, 2003.
 |
Summary |
|---|
 Top Summary IntroductionMethodsResultsDiscussionReferences |
|---|
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.
Keywords: intracerebral EEG; very fast oscillations; seizure onset; correlation
Abbreviations: SEEG= stereoelectroencephalography
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Low-voltage rapid discharges (very fast oscillations observed in ictal EEG activity) constitute a characteristic electrophysiological pattern in focal seizures of human epilepsy characterized by a noticeable decrease of signal voltage (sometimes preceded by high-amplitude spikes) 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 (depth-EEG) during pre-surgical evaluation of refractory partial epilepsies (Bancaud et al., 1973
). They occur at seizure onset and are often widespread.
Few neurophysiological studies have been dedicated to rapid low-voltage discharges compared with phasic interictal events (spikes and spike–waves). This may be explained by at least two facts. First, up to the last decade, high-frequency EEG activity was difficult to record with conventional EEG systems due to technical limitations (pen response times and low pass cutting frequency usually set at 
35 Hz in practice). Now, advances in digital EEG equipment allow signals to be sampled and recorded at very high frequency (typically 256 or 512 Hz in routine EEG, and up to 1 or 10 kHz for research purposes). Secondly, activity belonging to the gamma band of the EEG is also more difficult to observe in human scalp EEG recordings mainly due to contamination by muscle artefacts, especially at seizure onset with motor manifestations. Consequently, little information is available regarding high-frequency activity in clinical epilepsy. The accepted hypothesis is that rapid discharges often appear in brain regions in keeping with the epileptogenic zone (Bancaud, 1975), and that they might be a significant functional indicator of its spatio-temporal organization. The study of this electrophysiological pattern could thus provide relevant clues not only for the understanding of pathophysiological factors which initiate focal seizures in human but also for the definition of an optimal resection that could make the patient seizure free, as underlined by Alarcon et al. (1995) who analysed the power spectrum of ictal intracranial EEG signals and concluded that surgical removal of sites of localized high frequency could be associated with favourable prognosis.
In most of the studies dedicated to high-frequency oscillations observed during seizures, spectral analysis techniques were used to measure the power of EEG signals in defined frequency bands (typically delta, theta, alpha, beta and gamma). Ten years ago, Fisher et al. (1992) reported results from five patients suffering from different types of epilepsy. Fast Fourier transform power spectra revealed the presence of fast oscillations in two frequency bands (40–50 and 80–120 Hz). This activity was found to appear at seizure onset and often on ECoG (electrocorticography) contacts supposed to be close to the seizure focus. At the same time, Allen et al. (1992) reported high-frequency rhythmic activity in SEEG signals recorded in frontal lobe epilepsy. Autoregressive spectra showed that rapid activities may be observed in many sites over a wide region of the frontal lobe. Here again, these sites are supposed to be connected anatomically to the epileptogenic zone. More recently, Panzica et al. (1999) studied the spectral features of scalp EEG fast discharges during infantile spasms, and Mackenzie et al. (2002) reported results on the regional and spectral distribution of EEG rhythms in picrotoxin-induced seizures in the rat. In the former study, autoregressive analysis revealed the presence of a short rapid discharge of average frequency 20 Hz. In the latter, fast activity was found during seizures mainly in neocortex and in several subcortical areas.
However, to our knowledge, very few studies deal with spatio-temporal correlations in human intracerebral EEG signals measured during rapid discharges observed at seizure onset. Indeed, studies based on coherence analysis (Duckrow et al., 1992; Bartolomei et al., 1999; Zaveri et al., 1999) or on signal interdependencies (Repucci et al., 2001) do not deal specifically with ictal periods exhibiting high-frequency oscillations. However, studies related to the correlation of EEG signals specifically in the gamma band have been published, but have been conducted mainly during cognitive activation (Menon et al., 1996; Tallon-Baudry et al., 1998).
The present study focuses on spectral properties of rapid discharges observed at seizure onset and on the evolution of the correlation measured between SEEG signals generated from distant sites before, during and after such discharges. Signal analysis techniques aimed at estimating a linear correlation coefficient in typical EEG sub-bands and statistical tests reveal that signals are significantly de-correlated during ictal rapid discharges (reflected, at the EEG level, by a significant increase of power in the upper frequency band, mainly between 60 and 90 Hz) compared with periods that precede and that follow these discharges. The hypothesis of a functional decoupling of distant brain sites occurring at the start of seizures and preceding an abnormally high re-coupling when the seizure develops, formulated from the results obtained, is discussed.
|
Methods |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Patient selection and SEEG recording
Ten patients undergoing pre-surgical evaluation of drug-resistant partial epilepsy were selected for the presence of fast ictal activities in SEEG recordings performed during long-term video-EEG monitoring. SEEG was carried out as part of our patients’ normal clinical care, and they gave informed consent in the usual way. Our patients are informed that their data might be used for research purposes. All patients had a comprehensive evaluation including detailed history and neurological examination, neuropsychological testing, routine MRI study, scalp EEG and depth-EEG recording of seizures. Selection was dependent on the electrophysiological pattern at seizure onset (presence of rapid discharges revealed by visual analysis) and independent of the type of epilepsy (frontal or fronto-temporal). Table 1 provides clinical information about selected patients (age, sex, syndrome and aetiology) as well as the state of vigilance during analysed seizures.
|
SEEG recordings were performed using intracerebral multiple lead electrodes (10–15 leads; length, 2 mm, diameter, 0.8 mm; 1.5 mm apart) placed intracranially according to Talairach’s stereotactic method (Bancaud et al., 1973
; Talairach et al., 1974). The positioning of electrodes is determined in each patient from available non-invasive information and hypotheses about the localization of his/her epileptogenic zone (defined as the cerebral regions simultaneously discharging at seizure onset). The implantation accuracy is controlled peroperatively by telemetric X-ray imaging. A postoperative CT scan without contrast product is then used to verify both the absence of bleeding and the precise 3D location of each lead. Then, intracerebral electrodes are removed and an MRI is performed on which the trajectory of each electrode remains visible. Finally, a CT scan–MRI data fusion is performed to locate each lead anatomically along each electrode trajectory. All patients had electrodes positioned either in the prefrontal or frontal region. In some cases, the temporal region was also recorded (see below and Table 3). An example of intracerebral SEEG exploration is given in Fig. 1.
|
|
Signals were recorded on a DeltamedTM system on a maximum number of channels equal to 128. They were sampled at 256 Hz and recorded to hard disk (16 bits/sample) using no digital filter. The only filter present in the acquisition procedure is a hardware analogue high-pass filter (cut-off frequency equal to 0.16 Hz) used to remove non-physiological very slow variations that sometimes contaminate the baseline.
SEEG signal analysis
Bandpass filtering of SEEG signals and selection of distant sites exhibiting 
activity
The time series were decomposed on a filter bank, for which cut-off frequencies were chosen in accordance with classical EEG frequency sub-bands theta, alpha and beta. For the upper frequency band, we used a high-pass filter with a cut-off frequency equal to 24 Hz in order to detect the fastest oscillations, given our recording system. This band, denoted by the symbol 
, thus ranges from 24 to 128 Hz and is larger than the classical gamma band often denoted by the symbol and traditionally defined as ranging from 24 to 80 Hz (Niedermeyer and Lopes Da Silva, 1999).
The frequency range corresponding to each sub-band is given in Table 2. Hamming finite impulse response filters were chosen for their linear phase that is more appropriate for the computation of correlation coefficients (r2) in selected sub-bands than a non-linear phase (see next section). A filter order equal to 256 (1 s duration for the sampling rate of 256 Hz) that gave satisfactory impulse responses for each sub-band was retained. Readers may refer to Hamming (1998) for additional information about digital filter design. The signal average power PB was then estimated in each sub-band B over a 5 s long sliding window W stepped every 0.250 ms according to the following equation: PB = 
WsB2 (t)dt, where sB(t) denotes signal s(t) filtered in sub-band B.
|
For each patient, three distant brain sites (different electrodes) generating rapid discharges at seizure onset were selected from the estimation of the signal power P
in sub-band (highest frequencies). In order to facilitate this selection and to make it independent of visual choice, a colour-coded map, referred to as the activity map, was computed on one SEEG recording reflecting each patient’s typical seizure pattern. The activity map is a 2D representation of the evolution of P as a function of both time and space (i.e. the space of recording channels, up to 128). As detailed in Fig. 2, when computed on the seizure onset period and associated with the sorting of P values (descending order), activity maps allow cerebral regions that generate signals with maximal power in the band (fast activity) to be identified objectively.
|
Estimation of the spatio-temporal correlations of SEEG signals in sub-bands
For each one of the three pairs of signals built from the three SEEG signals (selected from the power in the
sub-band criterion, see previous section), spatial correlations were estimated from linear regression. This method can be used to compute a correlation coefficient, referred to as r2, that indicates to what extent samples contained in a temporal sliding window defined on two SEEG signals are correlated. Readers may refer to Pijn (1990) for details about methodological aspects of linear and non-linear regression analysis. The methodological novelty with respect to the aforementioned works is that we applied the method on the bandpass-filtered version of SEEG signals in order to evaluate interdependencies (amplitude–phase relationships) as a function of both time and frequency band. In order to limit the influence of the common reference in the computation of spatio-temporal correlation, bipolar signals obtained from subtraction of monopolar signals recorded on two adjacent leads were preferred to monopolar signals themselves. Readers may refer to Bullock et al. (1995) for discussion on the effect of a common reference on the measurement of cross-correlation. Briefly, the method proceeds in two steps. First, the two signals X and Y to be analysed are filtered, as described in the previous section. Secondly, the squared linear correlation coefficient r2 is estimated on each pair of filtered signals X and Y (i.e. in each sub-band) as a function of time. r2 values range from 0.0 (no linear relationship exists between signals X and Y in the considered frequency band) to +1.0 (a constant phase relationship exists between oscillations present in signals X and Y in the considered frequency band).
Coefficient r2 was computed on each pair of signals recorded from distant sites on the whole seizure duration. The analysis then focused on three distinct periods visually defined from the time of occurrence of the rapid discharge observed at seizure onset, as shown in Fig. 3. The first period (denoted ‘BD’ for ‘before discharge’) corresponds to the period preceding the discharge, the second one (denoted ‘D’ for ‘discharge’) corresponds to the discharge itself and the third period (denoted ‘AD’) corresponds to the period that follows the discharge when the frequency gradually decreases and when the spike activity of increasing amplitude starts to appear. Visual identification of the discharge period (D) as well as definition of starting and ending times are straightforward since this electrophysiological pattern is very characteristic (sudden amplitude decrease and frequency increase). Once the discharge period was defined, a 1 min time interval was chosen systematically for the BD period. The third period (AD) was defined as the time interval from the end of the discharge to the end of the seizure. For these three periods, both the power values per sub-band and the r2 values per sub-band were averaged over time. Finally, for each patient and for each period (BD, D or AD), the relative average power per sub-band was determined for each one of the three selected brain sites, and the average r2 per sub-band was determined for each one of the three distinct pairs of distant sites. To facilitate visual comparison, results were represented using histograms, as shown in Fig. 3E and F that give examples of relative average power and average r2 obtained from the application of the method on two SEEG signals recorded from two distant brain sites before, during and after a low-voltage rapid discharge.
|
Statistical analysis
Statistical analysis focused on r2 histograms in order to determine if there exist significant differences in the evolution of correlations between signals from distant sites which are computed over the pre-ictal/ictal periods (i.e. BD, D and AD periods), respectively. For each patient and for each pair of distant sites, r2 values per sub-band were first summed and normalized according to the following equation w = {ln[(1 + r)/(1 – r)]}/2 in order to make their distribution Gaussian (Bendat and Piersol, 1971
). Then, a Student t test and an ANOVA (analysis of variance) were performed in order to determine whether normalized distributions of correlation values over the BD, D and AD periods have the same, or significantly different, mean. |
Results |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
activity maps were computed on one seizure recording (typical pattern) per patient, as described above, and enabled the determination of three brain sites generating signals with high power in the upper frequency band at seizure onset. Results are given in Table 3. In all patients, frontal structures were involved during the ictal rapid discharge, except in patient GAU in whom fast activity mainly appeared in temporal structures (temporal lobe epilepsy). In some patients, distant sites may have been involved simultaneously during the initial discharge, as in patient POM in whom signals from the orbito-frontal cortex, the superior temporal gyrus and the lateral part of the temporal pole suddenly exhibited a low-voltage rapid discharge after a short period of spikes at the beginning of each seizure (note that these structures are thought to belong to the network characterizing the epileptogenic zone in this patient). The duration of ictal discharges measured in the 10 patients was found to vary from 5 to 17 s (mean: 9.3 s, SD: 3.4 s).
Signal average power and average r2 were then computed and represented as histograms for the three retained sites. Consequently, in each patient, a set of six histograms was produced: one histogram per SEEG signal displaying the relative average power per sub-band and one histogram per pair of SEEG signals displaying the average r2 value per sub-band. Visual analysis of the results shown in Fig. 4 reveals two major findings. First, for the signal power per sub-band, the distribution of power varies considerably as a function of the pre-ictal/ictal period. For most patients, before discharge (BD period), the average power was distributed mainly in the lowest frequency bands (
and 
). In some cases, significant power may also be observed in higher sub-bands (ß and ) often associated with a typical electrical pattern corresponding to nested rhythms (spikes in the band mixed with short rapid discharges in the ß/ band) that occur just before the low-voltage rapid discharge [patient TAL, signals recorded on electrodes TR, leads 7–8; and AC, leads 7–8 (see Fig. 1 legend for an explanation of electrode names and the numbering of electrode leads)]. During the discharge (D period), average power became dominant in sub-band . This is not surprising since signals were selected precisely on that criterion. Nevertheless, in some patients, the relative power in sub-band may be very high (patients POM, FO1–FO2; LEG, PA5–PA6; BEG, GP'14–GP'15, etc.) denoting a very tonic rapid discharge at the start of the seizure. In all patients, power spectral densities measured during the discharge period revealed that the very fast oscillations observed were located preferentially in a band from 60 to 90 Hz (see example in Fig. 2). After the discharge (AD period), the power distribution returned to more standard values with a strong decrease in sub-band .
|
Secondly, for correlation values, the most striking observation was related to the large decrease of the average r2 coefficient during the discharge. For all patients, the height of the second histogram column (corresponding to the discharge period) was far less than that of the two other columns, indicating a marked decorrelation between SEEG signals during fast activity (low-voltage rapid discharge). For patients in whom more than three distinct regions were involved in rapid ictal activity, computed values led to a similar observation. Moreover, in half of the patients, r2 values measured over the AD period were higher than those measured over the BD period, indicating that spatial correlations after the discharge may have strengthened once the rapid discharge vanished.
A statistical analysis of estimated average r2 values was then conducted in order to determine whether their distribution over the three periods differed significantly. Statistical results are presented in Fig. 5. First, r2 values were summed over the frequency band in order to build a global indicator of the spatial correlation (independent of the frequency band). Figure 5A displays the 30 summed r2 values obtained in the 10 patients for the three retained pairs of signals. As visually determined from the histograms (Fig. 4), values corresponding to the discharge period were lower than those estimated over the other two periods, and values corresponding to the BD period seemed lower than those obtained for the AD period. In order to test these observed mean differences statistically, values were normalized, as described above. The empirical distribution of normalized r2 values, shown in Fig. 5B, may be assumed to be Gaussian. A conventional t test was then performed to measure the significance of observed differences of the mean. As shown in Fig. 5D, the small numerical value of the significance (P value) is <0.01, indicating that measured correlations during the D period are statistically lower than those measured during the other two periods. The same also applies for the comparison of values obtained during the BD and AD periods. Finally, a one-way ANOVA was performed using a standard tool of numerical analysis (Matlab’s ANOVA routine, statistics toolbox). The results, displayed in Fig. 5C in the form of a so-called ‘boxplot’, confirm those obtained with the t test. The lower and upper lines of each ‘box’ (25th and 75th percentiles of each sample) and the line in the middle of the box (sample median) indicate, respectively, that the distribution of r2 values during the discharge (D period) was narrower and that its mean was lower than the two other distributions. The ‘whiskers’ (lines extending above and below the box) show the extent of the rest of the sample. Of interest is the presence of outliers (defined as values that are >1.5 times the interquartile range away from the top or bottom of the box) in the three distributions (two outliers for the BD period, three outliers for the D period and one outlier for the AD period). Note that, in patient TAL, correlation values measured before discharge (BD) were significantly higher than those measured in the other patients. In the same patient, two correlation values measured during the discharge (D) were lower than those (already low) measured in other patients. The opposite was found in patient ALB, where one correlation value was slightly higher than those obtained elsewhere, meaning that for one pair of distinct sites (PM5–PM6 versus FP9–FP10), decorrelation was less marked, but still remained statistically lower than those obtained on the BD and AD periods, as in other patients. Finally, in patient GAU, a statistically higher correlation value was observed over the AD period. In summary, for patients TAL and GAU, the presence of outliers denotes a desynchronization/synchronization mechanism more marked than that observed in other patients.
|
The notches in boxes graphically show the confidence interval (95%) about the median of each distribution. The fact that they do not intersect indicates that means differed significantly from period to period: r2 values measured during the discharge were lower than those measured before and after, and values measured after the discharge were higher than those measured before discharge.
|
Discussion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Results show that correlation between SEEG signals recorded from distinct neural structures and exhibiting a low-voltage rapid discharge (very fast oscillations) at seizure onset is dramatically lower than the correlation measured over periods that come before and after this discharge. This observation deserves to be discussed from several viewpoints.
First, from the signal analysis viewpoint, spatial correlations (in a general sense) estimated using various available methods (coherence, linear and non-linear regression, mutual entropy, mutual prediction, etc.) can be related to functional couplings between brain sites that generate signals (paradigm). For the two last decades, this type of analysis led to relevant results not only in the field of epilepsy but also in the field of perception/cognition. Along these lines, the observed decorrelation reported in this study would be interpreted as an abnormal decoupling between recorded distant sites at seizure onset. To our knowledge, this concept of ‘disruption’ has received less attention than that of ‘hyperactivation’ revealed by significantly high correlations between signals recorded in the epileptogenic zone (Bartolomei et al., 2002). Both concepts could be operative in analysing pathophysiology in human epileptic seizures.
Secondly, in all patients, the rapid ictal discharge involves distant and functionally distinct brain sites almost simultaneously, showing that a ‘synchronizing system’ may exist which gives rise to the simultaneous start of fast activity. After a few hundred milliseconds, once the rapid discharge is running, spatial correlation significantly decreases as if brain sites involved in very fast oscillations were being desynchronized. This observation refers back to old debates about EEG correlates of sleep–waking cycles. Desynchronization associated with arousal was thought to result from the activation of ascending reticular projections to the cortex. By analogy with this physiological situation, Jasper interpreted EEG flattening (closely related to low-voltage rapid discharge as discussed in Bancaud et al., 1973) as conveying neuronal desynchronization. However, no basic mechanism of epilepsy available at that time supported this hypothesis. Current concepts were based on hyperexcitability, paroxysmal depolarizing shift (PDS) and hypersynchrony on the one hand, and alteration of inhibition on the other hand (Delgado-Escueta et al., 1999). However, the question of synchronization/desynchronization of larger neuronal pools was not the object of the basic studies. Therefore, the present study again puts desynchronization as a central question.
Thirdly, from a neurophysiological viewpoint, we noticed that very fast oscillations observed at seizure onset are located mainly in the 60–90 Hz band, which is actually higher than the EEG gamma band, classically defined as between 24 and 80 Hz. This may lead to the establishment of a link between observed fast oscillations and two other types of fast oscillations, respectively termed ‘ripples’ (80–200 Hz, occurring during normal processes) and ‘fast ripples’ (200–500 Hz, occurring in epileptogenic tissue). Indeed, although the duration of observed fast oscillations (5–15 s) at seizure onset is far higher than that of ripples (a few hundred milliseconds), both share common features. First, a recent study by Grenier et al. (2001) shows that ripples (i) may appear almost simultaneously in different cortical sites and (ii) are uncorrelated among different gyri (but correlated among sites within the same gyrus). Secondly, results from Bragin et al. (1999) suggest that the fast ripple is unique to tissue capable of generating seizures, in human epileptic brain and in kainic acid-treated rats. Of course, we are unable to determine whether such oscillations are also present in our observations since the sampling frequency of our recording system is only 256 Hz. However, in both cases, recordings are performed from epileptogenic regions. Thirdly, the aforementioned studies and several other studies show that inhibitory interneurons may play a role in the patterning of neocortical ripples as well as in the generation of epileptic very fast oscillations. For example, results described in Penttonen et al. (1998) suggest that rapid discharges reflect rhythmic postsynaptic potentials in pyramidal cells brought about by rhythmically discharging somatic-projecting inhibitory interneurons. According to the authors’ in vivo studies, a few presynaptic inhibitory interneurons are sufficient to impose a gamma rhythm on pyramidal neurons and time the occurrence of their action potentials. We also reported similar results on high-frequency EEG activities using a new computational macroscopic model of EEG activity that includes a physiologically relevant fast inhibitory feedback loop representing the role of somatic-projecting interneurons (Wendling et al., 2002). We showed that strikingly realistic fast oscillations are produced by the model when compared with real depth-EEG epileptic signals recorded with intracerebral electrodes. Moreover, in a comprehensive study of spike–wave complexes and fast components of cortically generated seizures (a series of four papers), Steriade and Contreras (1998) reported corroborative data on fast runs (that are not exactly in the gamma band since their frequency range is 12–15 Hz) observed at the onset of cortical seizures. These authors emphasize that (i) short-lasting inhibitory processes may survive during fast runs as shown by inhibitory postsynaptic potentials (IPSPs) occurring during the depolarizing plateau in which paroxysmal fast runs appear and action potentials are inactivated; and (ii) intracortical excitation overwhelms inhibitory processes, which is an effective factor in depolarizing large populations of cortical neurons. From these results, we can thus hypothesize that the observed signal decorrelation during fast activity may be the consequence of a local sustained discharge of inhibitory interneurons onto pyramidal cells, reflected, at the EEG level, by an activity located in a relatively wide high-frequency band and spatially decorrelated from one site to another.
Another question arises from the results: how does epileptic fast activity relate to cognitive gamma activity? To us, very fast oscillations encountered in low-voltage rapid discharges constitute a characteristic electrophysiological pattern of human partial epilepsy and are pathological, as opposed to gamma oscillations recorded during normal cognitive tasks. The first argument is related to frequency. As mentioned above, the frequency band of observed epileptic fast oscillations ranges from 60 to 90 Hz. These values are actually higher than those classically associated with cognitive gamma activity. The second argument is that recorded epileptic rapid discharges may last up to 10 s, which is not the case for cognitive gamma. This point is also discussed in Traub (1999) who gives a classification of types of in vivo gamma oscillations as spontaneous, induced, evoked, emitted and epileptic. He also presents another argument: epileptic gamma oscillations are encountered in cerebral structures that do not normally generate this type of activity. For example, gamma frequency population spikes have not been recognized in vivo in the hippocampus as a spontaneously occurring event, as opposed to following repetitive stimulation or drug application. This point is particularly important because the main difference between normal gamma oscillations and epileptic very fast oscillations could reside in the fact that the former are coherent or synchronous (between involved brain areas) whereas the latter are uncoherent or desynchronized.
Finally, we will discuss the reported observation with respect to clinical signs that are observed in the patient behaviour at the start of and during seizures involving rapid discharges. The fact that the observed signal decorrelation can be related to a sudden functional decoupling that occurs at seizure onset may influence the way in which we interpret clinical semiology. Indeed, coherent fast activity has been postulated to be a way for the brain to process and integrate information. Any interruption of this coherence would then result in the break-up of cerebral cortex processing, leading, for instance, to altered cognition and/or behaviour as observed in epileptic patients during seizures. A related concept has been proposed for the dreamy state by Jackson (1931), who postulated three hierarchical levels of consciousness, progressively impaired according to seizure spread. In such a conception, the fast discharge would lead to a loss of highest functions at the highest level, releasing the second level, which is physical consciousness and which produces the dreamy state: "the disease places ‘hors de combat’" certain nervous elements or regions of the highest centres so that "not only illusions but all positive mental symptoms [...] are the outcome of activity of lower, non-diseased nervous arrangements of these highest centres". We suggest that Jackson’s conception might certainly be extended to other types of symptoms and signs. Consequently, clinical symptoms such as automatisms or forced acting would also correspond to the activation of centres liberated from the control of the highest centres impaired by fast epileptic activities.
|
References |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Alarcon G, Binnie CD, Elwes RD, Polkey CE. Power spectrum and intracranial EEG patterns at seizure onset in partial epilepsy, Electroencephalogr Clin Neurophysiol 1995; 94: 326–37.[CrossRef][Web of Science][Medline]
Allen PJ, Fish DR, Smith SJ. Very high-frequency rhythmic activity during SEEG suppression in frontal lobe epilepsy. Electroencephalogr Clin Neurophysiol 1992; 82: 155–9.[CrossRef][Web of Science][Medline]
Bancaud J. Stereoelectroencephalography. In: Remond A, editor. Handbook of electroencephalography and clinical neurophysiology, Vol. 10(B). Amsterdam: Elsevier; 1975. p. 10(B)3–65.
Bancaud J, Talairach J, Geier S, Scarabin JM. EEG et SEEG dans les tumeurs cérébrales et l’épilepsie. Paris: Ed. Edifor; 1973.
Bartolomei F, Wendling F, Vignal JP, Kochen S, Bellanger JJ, Badier JM, et al. Seizures of temporal lobe epilepsy: identification of subtypes by coherence analysis using stereo-electro-encephalography. Clin Neurophysiol 1999; 110: 1741–54.[CrossRef][Web of Science][Medline]
Bartolomei F, Wendling F, Vignal JP, Chauvel P, Liégeois-Chauvel C. Neural networks underlying epileptic humming, Epilepsia 2002; 43: 1001–12.[CrossRef][Web of Science][Medline]
Bendat JS, Piersol AG. Random data: analysis and measurement procedures. New York: Wiley-Interscience; 1971.
Bragin A, Engel J Jr, Wilson CL, Fried I, Mathern GW, Hippocampal and entorhinal cortex high-frequency oscillations (100–500 Hz) in human epileptic brain and in kainic acid-treated rats with chronic seizures. Epilepsia 1999; 40: 127–37.[CrossRef][Web of Science][Medline]
Bullock TH, McClune MC, Achimowicz JZ, Iragui-Madoz VJ, Duckrow RB, Spencer SS. EEG coherence has structure in the millimeter domain: subdural and hippocampal recordings from epileptic patients. Electroencephalogr Clin Neurophysiol 1995; 95: 161–77.[CrossRef][Web of Science][Medline]
Delgado-Escueta AV, Wilson WA, Olsen RW, Porter RJ, Pietsch SG, editors. Jasper’s basic mechanisms of the epilepsies, 3rd edn. Advances in neurology, Vol. 79. Philadelphia: Lippincott Williams &Wilkins; 1999.
Duckrow RB, Spencer SS. Regional coherence and the transfer of ictal activity during seizure onset in the medial temporal lobe. Electroencephalogr Clin Neurophysiol 1992; 82: 415–22.[CrossRef][Web of Science][Medline]
Fisher RS, Webber WR, Lesser RP, Arroyo S, Uematsu S. High-frequency EEG activity at the start of seizures. J Clin Neurophysiol 1992; 9: 441–8.[Web of Science][Medline]
Grenier F, Timofeev I, Steriade M. Focal synchronization of ripples (80–200 Hz) in neocortex and their neuronal correlates. J Neurophysiol 2001; 86: 1884–98.
Hamming RW. Digital filters, 3rd edn. Mineola (NY): Dover Publications; 1998.
Jackson JH. Selected writings of John Hughlings Jackson. Edited by Taylor J, Holmes G, Walshe FM. London: Hodder & Stoughton; 1931.
Mackenzie L, Medvedev A, Hiscock JJ, Pope KJ, Willoughby JO. Picrotoxin-induced generalised convulsive seizure in rat: changes in regional distribution and frequency of the power of electroencephalogram rhythms. Clin Neurophysiol 2002; 113: 586–96.[Medline]
Menon V, Freeman WJ, Cutillo BA, Desmond JE, Ward MF, Bressler SL, et al. Spatio-temporal correlations in human gamma band electrocorticograms, Electroencephalogr Clin Neurophysiol 1996; 98: 89–102.[CrossRef][Web of Science][Medline]
Niedermeyer E, Lopes Da Silva F. Electroencephalography: basic principles, clinical applications, and related fields, 4th edn. Baltimore: Williams & Wilkins; 1999.
Panzica F, Franceschetti S, Binelli S, Canafoglia L, Granata T, Avanzini G. Spectral properties of EEG fast activity ictal discharges associated with infantile spasms. Clin Neurophysiol 1999; 110: 593–603.[CrossRef][Web of Science][Medline]
Penttonen M, Kamondi A, Acsady L, Buzsaki G. Gamma frequency oscillation in the hippocampus of the rat: intracellular analysis in vivo. Eur J Neurosci 1998; 10: 718–28.[CrossRef][Web of Science][Medline]
Pijn JP. Quantitative evaluation of EEG signals in epilepsy, nonlinear associations, time delays and nonlinear dynamics [PhD Thesis]. Amsterdam: University of Amsterdam; 1990.
Repucci MA, Schiff ND, Victor JD. General strategy for hierarchical decomposition of multivariate time series: implications for temporal lobe seizures. Ann Biomed Eng 2001; 29: 1135–49.[Medline]
Steriade M, Contreras D. Spike–wave complexes and fast components of cortically generated seizures, I to IV. J Neurophysiol 1998; 80: 1439–513.
Talairach J, Bancaud J, Szickla G, Bonis A, Geier S. Approche nouvelle de la chirurgie de l’épilepsie: methodologie stéréotaxique et résultats thérapeutiques. Neurochirurgie 1974; 20 Suppl 1: 1–240.[Medline]
Tallon-Baudry C, Bertrand O, Peronnet F, Pernier J. Induced gamma-band activity during the delay of a visual short-term memory task in humans. J Neurosci 1998; 18: 4244–54.
Traub RD, Jefferys JGR, Whittington MA. Fast oscillations in cortical circuits. Cambridge (MA): MIT Press; 1999.
Wendling F, Bartolomei F, Bellanger JJ, Chauvel P. Epileptic fast activity can be explained by a model of impaired GABAergic dendritic inhibition. Eur J Neurosci 2002; 15: 1499–508.[CrossRef][Web of Science][Medline]
Zaveri HP, Williams WJ, Sackellares JC, Beydoun A, Duckrow RB, Spencer SS. Measuring the coherence of intracranial electroencephalograms. Clin Neurophysiol 1999; 110: 1717–25.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  F. Wendling, F. Bartolomei, and L. Senhadji Spatial analysis of intracerebral electroencephalographic signals in the time and frequency domain: identification of epileptogenic networks in partial epilepsy Phil Trans R Soc A, January 28, 2009; 367(1887): 297 - 316. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Bartolomei, P. Chauvel, and F. Wendling Epileptogenicity of brain structures in human temporal lobe epilepsy: a quantified study from intracerebral EEG Brain, July 1, 2008; 131(7): 1818 - 1830. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. McGonigal, F. Bartolomei, J. Regis, M. Guye, M. Gavaret, A. T.-D. Fonseca, H. Dufour, D. Figarella-Branger, N. Girard, J.-C. Peragut, et al. Stereoelectroencephalography in presurgical assessment of MRI-negative epilepsy Brain, December 1, 2007; 130(12): 3169 - 3183. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Bartolomei, M. Guye, F. Wendling, A. McGonigal, J. Regis, and P. Chauvel Reply: On the Role of Medial Pulvinar Thalamic (PuM) Nucleus in Temporal Lobe Epilepsy Brain, May 1, 2007; 130(5): e72 - e72. [Full Text] [PDF]
|
|
|
|
|
|
F. Mormann, R. G. Andrzejak, C. E. Elger, and K. Lehnertz Seizure prediction: the long and winding road Brain, February 1, 2007; 130(2): 314 - 333. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Schindler, H. Leung, C. E. Elger, and K. Lehnertz Assessing seizure dynamics by analysing the correlation structure of multichannel intracranial EEG Brain, January 1, 2007; 130(1): 65 - 77. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||