Skip to main content
NCBI home page
PMC Canada Author Manuscripts logoLink to PMC Canada Author Manuscripts
. Author manuscript; available in PMC: 2013 Aug 27.
Published in final edited form as: Epilepsia. 2012 Mar 14;53(5):797–806. doi: 10.1111/j.1528-1167.2012.03428.x

Continuous high-frequency activity in mesial temporal lobe structures

Francesco Mari 1, Rina Zelmann 1, Luciana Andrade-Valenca 1, Francois Dubeau 1, Jean Gotman 1
PMCID: PMC3753292  CAMSID: CAMS3340  PMID: 22416973

Summary

Purpose

Many recent studies have reported the importance of high-frequency oscillations (HFOs) in the intracerebral electroencephalography (EEG) of patients with epilepsy. These HFOs have been defined as events that stand out from the background. We have noticed that this background often consists itself of high-frequency rhythmic activity. The purpose of this study is to perform a first evaluation of the characteristics of high-frequency continuous or semicontinuous background activity.

Methods

Because the continuous high-frequency pattern was noted mainly in mesial temporal structures, we reviewed the EEG studies from these structures in 24 unselected patients with electrodes implanted in these regions. Sections of background away from interictal spikes were marked visually during periods of slow-wave sleep and wakefulness. They were then high-passed filtered at 80 Hz and categorized as having high-frequency rhythmic activity in one of three patterns: continuous/semicontinuous, irregular, sporadic. Wavelet entropy, which measures the degree of rhythmicity of a signal, was calculated for the marked background sections.

Key Findings

Ninety-six bipolar channels were analyzed. The continuous/semicontinuous pattern was found frequently (29/96 channels during wake and 34/96 during sleep). The different patterns were consistent between sleep and wakefulness. The continuous/semicontinuous pattern was found significantly more often in the hippocampus than in the parahippocampal gyrus and was rarely found in the amygdala. The types of pattern were not influenced by whether a channel was within the seizure-onset zone, or whether it was a lesional channel. The continuous/semicontinuous pattern was associated with a higher frequency of spikes and with high rates of ripples and fast ripples.

Significance

It appears that high-frequency activity (above 80 Hz) does not appear only in the form of brief paroxysmal events but also in the form of continuous rhythmic activity or very long bursts. In this study limited to mesial temporal structures, we found a clear anatomic preference for the hippocampus. Although associated with spikes and with distinct HFOs, this pattern was not clearly associated with the seizure-onset zone. Future studies will need to evaluate systematically the presence of this pattern, as it may have a pathophysiologic significance and it will also have an important influence on the very definition of HFOs.

Keywords: High-frequency EEG, Mesial temporal structures, High-frequency oscillations, Intracerebral electrodes

High-frequency oscillations (HFOs) are brief and transient events that occur in conjunction with interictal spikes or independently and, at high rates, are considered reliable markers of epileptogenicity (Bragin et al., 2010; Gotman, 2010; Jacobs et al., 2010). Two types of events are distinguished: ripples (80–250 Hz) and fast ripples (FRs, 250–500 Hz). Whereas 80–200 Hz oscillations are also related to physiologic activity (memory or sensory processing, Axmacher et al., 2008; Nagasawa et al., 2012), higher frequencies, between 200 and 500 Hz, seem to be more closely linked with epileptogenic tissue (Bragin et al., 2002). HFOs were first studied in humans using microwires (with a surface contact of 70 μm2) and were found within epileptic tissue and outside in the less-affected hippocampus of patients with epilepsy studied with bilateral temporal lobe implantation (Bragin et al., 1999a,b; Staba et al., 2002, 2004). A series of studies demonstrated the possibility of detecting this high-frequency activity with clinical macro-electrodes (surface area of 1 mm2) and demonstrated that interictal HFOs are strongly associated with the seizure-onset zone (SOZ) (Urrestarazu et al., 2007; Jacobs et al., 2008), could mirror the disease activity (Zijlmans et al., 2009), and could be used as a marker of the epileptogenic zone (Jacobs et al., 2010; Wu et al., 2010). HFOs were first and most often described in the mesial temporal lobe structures (Bragin et al., 1999a; Staba et al., 2002; Worrell et al., 2008), but then also found in neocortical regions, although at generally lower rates (Urrestarazu et al., 2007; Jacobs et al., 2008). We observed, in some patients with electrodes in mesial temporal structures, a particular pattern characterized by a continuous or semicontinuous high-frequency activity in the ripple frequency range. Brief periods of higher amplitude activity were labeled “ripples” as they stood out from the background, but the background itself appeared constituted of rhythmic activity.

We evaluated this pattern in 24 patients with electrodes in mesial temporal lobe structures, looking for its incidence and for a correlation between this pattern and different electroclinical and neuroradiologic variables.

Methods

Patients

Between October 2004 and October 2007, 42 patients were recorded with intracranial electrodes in the Epilepsy Unit of the Montreal Neurological Institute. The stereo–electroencephalography (SEEG) studies were performed when comprehensive noninvasive presurgical evaluation yielded inconclusive results. The sites of electrode placement were individualized according to clinical history, seizure semiology, neuroimaging, and surface EEG investigations. We selected EEG sections during wakefulness and slow-wave sleep in the 26 patients with electrodes in mesial temporal lobe structures (amygdala/uncus, Am; anterior hippocampus, Hp; and middle hippocampus/parahippocampal gyrus, PH) recorded consecutively in this period and who had not had previous surgery.

Recording methods

SEEGs were recorded using the Harmonie monitoring system (Stellate, Montr3al, QC, Canada). Electrode bundles were implanted stereotactically using an image-guidance system (SSN Neuronavigation System, Mississauga, ON, Canada) through percutaneous holes drilled in the skull. Typically, depth electrodes in the temporal lobe were directed orthogonally through the middle temporal gyrus in anterior, mid, and posterior locations such that the deepest contacts became situated in Am, Hp, and PH, respectively. Intracerebral electrodes were manufactured on-site by wrapping 3/1,000 inch stainless steel wire around a 10/1,000 inch stainless steel central core. These wires were coated with Teflon, except for regions where the insulation was stripped to form electrode contacts. Each electrode had nine contacts, with the deepest contact (contact 1) consisting of the tip of the steel core stripped of insulation. This contact had a length of 1 mm, whereas all other contacts (2–9) were formed from stripped sections of the marginal wire that was tightly wound to create 0.5-mm–long coils. The effective surface area was 0.80 mm2 for contact 1 and 0.85 mm2 for contacts 2–9. At times, SEEG was acquired with a 500-Hz low-pass hardware filter and a sampling rate of 2,000 Hz (other times 70-Hz, filter, 200-Hz sampling rate). Only sections recorded at 2,000 Hz with coregistered electromyography–electrooculography (EMG-EOG) channels were selected for the study. SEEG analysis was performed in bipolar montages on interictal samples of wakefulness and slow-wave sleep. The epochs were selected at least 2 h before and after a seizure, to reduce the influence of seizures in our analysis. Wakefulness epochs lasting 5 min were selected from the first daytime recording. Slow-wave sleep epochs lasting 10 min were selected from the first nighttime recording (see below the rationale for using different durations in wake and sleep). To confirm the sleep stage, the Harmonie software was used to compute spectral trends in the delta, alpha, and beta bands for the intracranial EEG and power of the chin EMG with a 30-s time resolution. The EEG sections with high delta and low EMG power were visually reviewed and selected as slow-wave sleep.

Contacts selection and classification

We systematically chose the deepest two pairs of contacts assuming that they are located in hippocampal, parahippocampal, and amygdala gray structures. We classified these contacts as lesional or nonlesional by combining the information from the reconstructed positions of the electrodes from the Neuronavigation system, a computed tomography (CT) obtained directly after the implantation, and a magnetic resonance imaging (MRI) after explantation. Lesional borders were defined by the neurosurgeon and neuroradiologist after examination of all available clinical MRI studies (global T1 with gadolinium, T1 sagittal, T2 axial and coronal, and fluid-attenuated inversion recovery [FLAIR] coronal). For the purpose of this analysis, hippocampal atrophy was considered a lesion. Moreover, we define the contacts as inside or outside the seizure-onset zone (SOZ and nSOZ) if they displayed ictal activity (as identified by the clinical neurophysiologist who was reviewing the EEG for clinical purposes) at the beginning or during the first 5 s of the seizure-related electrical changes. In patients with two independent SOZs, we divided the SOZ contacts into SOZ and secondary-SOZ (sSOZ) depending on the side with the most seizures.

Visual analysis

Earlier studies have demonstrated that high-frequency activity is often suppressed immediately after spikes (Urrestarazu et al., 2006; Kobayashi et al., 2009; Jacobs et al., 2011). To avoid this influence of spikes, we analyzed only sections of SEEG separated from a spike by at least 1 s (Fig. 1A,B). Because spikes are more frequent during sleep, we analyzed longer sleep sections (10 min) than wake sections (5 min) to obtain sufficient interspike data. These interspike sections will be labeled “background.” Selection and analysis of background epochs involved the following steps.

Figure 1.

Figure 1

Open in a new tab

(A) Deepest pair of contacts of an intracranial electrode aimed at the Hp in a patient with left Hp atrophy. The unfiltered SEEG shows high-voltage spikes; the red section highlights an example of background selected for the study (defined as interspike SEEG segment spaced at least 1 s before and after the spikes); the thin green section is expanded in time and amplitude in the broad section below (750-msec duration) in which a C/SC oscillatory activity is shown. (B) Example of attenuation of high-frequency activity immediately following a spike, as demonstrated in Urrestarazu et al. (2006). (C) Deepest pair of contacts of the intracranial electrode aimed at the right Hp in a patient with right Hp atrophy: the SEEG shows extremely high spike rates that makes it impossible to detect background epochs.

Epilepsia © ILAE

Spike marking and background epoch selection

In wakefulness and sleep, SEEGs were reviewed visually to detect interictal spikes. All spikes were identified within the selected sample using common filter setting and time scale (filter 0.3–70 Hz, time scale 15 s/page). Background epochs were all sections outside of the 2-s epoch centered at each spike. One patient had to be excluded because of almost continuous spiking, leaving 25 patients (Fig. 1C).

Background classification

Channels were displayed with the maximum time resolution (0.92 s/page, 529.2 mm/s). The EEG was high-pass filtered at 80 Hz using a finite impulse response filter to minimize ringing. Epochs were classified visually in three patterns depending on the length of the oscillations and on the presence of a clear separation between the transient elements (Fig. 2): (1) Continuous/Semicontinuous (C/SC): When the background is almost completely occupied by a continuous or semicontinuous high-frequency oscillatory activity (oscillatory activities lasting at least 500 ms and separated by <100 ms); (2) Irregular (IRR): the background activity is characterized by the presence of HFO lasting >200 msec; if the duration is >500 msec, then oscillations must be separated by more than 100 ms; (3) Sporadic (SP): when the background is neither C/SC nor IRR; this usually corresponds to infrequent short-duration oscillations. The values 500, 200, and 100 msec are relatively arbitrary; 500 msec attempts to capture a situation clearly different from the usual short ripples and fast ripples; 200 msec is meant as an intermediate value. Examples of these patterns are shown on Fig. 2, with the corresponding time frequency analysis for frequencies above 50 Hz. The figure also provides the value of wavelet entropy for each segment, a measure of the presence of rhythmic activity in the signal, which is independent of amplitude and frequency. The wavelet entropy (Rosso et al., 2001) was calculated after band-pass filtering the EEG (80–450 Hz) with a finite impulse response filter and computing the autocorrelation of the filtered signal. The autocorrelation was used to enhance the oscillatory behavior of the signal (Chander, 2007). One patient had to be excluded because of an artifactual continuous high frequency activity around 120 Hz probably caused by mains power interference. This left valid data from 24 patients.

Figure 2.

Figure 2

Open in a new tab

Examples of different background patterns. Each column presents a set of examples. For each graph the upper trace shows the original EEG; the second trace shows the signal high-pass filtered at 80 Hz; below this trace is the time-frequency analysis for this EEG segment, including the value of wavelet entropy (WE) for the segment (higher values of WE correspond to less rhythmic signal, independently of amplitude). (A) Continuous/semicontinuous; (B) Irregular; (C) Sporadic; (D) same signal as in (C) but with amplitude gain increased by factor of 5 to illustrate that the absence of rhythmic activity is not dependent on the gain of the display.

Epilepsia © ILAE

Marking ripples and FRs

We marked ripples and FRs according to the rules used in previous studies (Jacobs et al., 2009). HFOs were marked if the oscillation contained at least four consecutive waves and was clearly visible above the background signal in filtered data (with the need of a separation between events lasting the equivalent of at least two oscillations). In the case of the C/SC pattern, HFOs were rhythmic events of amplitude clearly larger than the continuous rhythmic activity of the background. Examples are shown on Fig. 3.

Figure 3.

Figure 3

Open in a new tab

Examples (green sections) of HFOs visually marked in channels displaying the C/SC type of background activity (oscillations with at least four waves, larger amplitude than the background, sinusoidal shape).

Epilepsia © ILAE

Statistical analysis

After marking all events, a MATLAB program (Math-Works, Natick, MA, U.S.A.) calculated the spike, ripple, and FR rates per minute in the slow wave sleep and wakeful-ness epochs selected. Given the lack of Gaussian nature of the rate distributions, we used nonparametric tests and reported median rates values for all analyses. Because background patterns seem to have an order in the degree of oscillatory behavior (with C/SC being the most oscillatory, followed by IRR, and SP being the most random background type), for some analysis, we considered the patterns as ordinal data. Therefore, Wilcoxon signed-rank test was used to determine if there were differences in background patterns as well as in spike/ripple/fast ripple rates between wakeful-ness and sleep. The Kruskal-Wallis one-way analysis of variance method (KW test) was used to compare the spike/ripple/FR median rates in the different background patterns and anatomic electrode locations. Statistical results were corrected, if necessary, for multiple comparisons and were analyzed with post hoc Tukey HSD (honestly significant differences). Chi-square test was used to compare background patterns with anatomic location, SOZ, and lesions. Spearman correlation was used to analyze the relation between spike, ripple, and FR rates and the wavelet entropy of each channel. The level of significance was set at 0.05 for all tests.

Results

In total, 96 mesial temporal lobe channels were evaluated in 24 patients (15 male, mean age standard deviation [SD], 35.2 11.5 years). Thirty-six contacts were in Am, 35 in Hp, and 25 in PH. The main anatomoelectroclinical characteristics of the patients are displayed in Table 1.

Table 1.

Main electroclinical characteristics of the patients

Pt Age/gender MRI findings SEEG implantation sites SOZ channels sSOZ channels Lesional channels
1 45/M Normal (5) L-TPo, L-Am, L-Hp, L-PH, L-OF L-Am, L-Hp, L-PH – –
2 34/F Normal (5) R-Am, R-Hp, R-PH, R-OF, R-ACi R-Am, R-Hp, R-PH – –
3 42/F R MTL atrophy (6) L-Am, L-Hp, L-PH, R-Am, R-Hp, R-PH R-Am, R-Hp, R-PH L-Am, L-Hp, L-PH R-Am, R-Hp, R-PH
4 30/M Normal (6) L-Am, L-Hp, R-Am, R-Hp, R-PH, R-I R-Am, R-Hp, R-PH L-Am, L-Hp –
5 37/M Large L CP porencephalic cyst (7) L-TP, L-Am, L-OF, L-OP, L-ACi, L-PO Not in MTL contacts – Not in MTL contacts
6 49/F R MTL atrophy (6) L-Am, L-Hp, L-PH, R-Am, R-Hp, R-PH R-Am, R-Hp, R-PH L-Am, L-Hp, L-PH R-Am, R-Hp, R-PH
7 46/M R Hp malrotation and atrophy (4) L-Am, L-Hp, R-Am, R-Hp R-Am, R-Hp – R-Am, R-Hp
8 42/M Normal (5) L-TPo, L-Am, L-Hp, L-PH, L-OF L-Am, L-Hp, L-PH – –
9 47/M Bilateral MTL atrophy (6) L-Am, L-Hp, L-PH, R-Am, R-Hp, R-PH R-Am, R-Hp, R-PH L-Am, L-Hp, L-PH R-Am, R-Hp, R-PH, L-Am, L-Hp, L-PH
10 45/F Normal (6) L-Am, L-Hp, L-PH, R-Am, R-Hp, R-PH L-Am, L-Hp, L-PH R-Am, R-Hp, R-PH –
11 30/M Normal (10) R-Am, R-Hp, R-SMA, R-OF, R-ACi, L-Am, L-Hp, L-SMA, L-OF, L-ACi Clear ictal changes after clinical SOZ – –
12 26/F Enlargement of R lateral ventricle (R hemiatrophy, HHE syndrome) (4) R-Am, R-Hp, R-PH, R Heschl gyrus RAm, RHp, RPH – Unclear
13 56/F L Me and NC T atrophy (6) L-Am, L-Hp, L-PH, R-Am, R-Hp, R-PH Clear ictal changes after clinical SOZ – L-Am, L-Hp, L-PH
14 17/M Normal (6) L-Am, L-Hp, L-PH, L-OF, L-ACi, L-MCi L-Am, L-Hp, L-PH – –
15 34/M Tuberous sclerosis complex (tuber in L F and T and R F and P) (8) L-OF, L-ACi, L-Am, L-Hp, R-OF, R-ACi, R-Am, R-Hp Not in MTL contacts – Not in MTL contacts
16 48/F Normal (4) R-Am, R-Hp, R-PH, R-Isthmus Clear ictal changes after clinical SOZ – –
17 25/M PNH (two nodular Het in R trigonal area) (5) R-Am, R-Hp, R-PH, R-ANod, R-PNod R-Am, R-Hp, R-PH – Not in MTL contacts
18 20/M Normal (7) R-Am, R-Hp, R-ACi, R-PCi, R-OF, R-IC, R-SC Not in MTL contacts – –
19 18/M Normal (6) L-Am, L-Hp, L-PH, R-Am, R-Hp, R-PH R-Am, R-Hp, R-PH L-Am, L-Hp, L-PH –
20 23/M Normal (6) R-Am, R-Hp, R-OF, R-GRe, R-ACi, R-SMA R-Am, R-Hp – –
21 44/M Normal (6) L-Am, L-Hp, L-PH, R-Am, R-Hp, R-PH L-Am, L-Hp, L-PH – –
22 43/F Multiple cavernomas (R F and L P) (5) R-Am, R-Hp, R-PH, R-ACi, R-SMA R-Am, R-Hp, R-PH – Not in MTL contacts
23 25/M Bilateral PNH (9) L-Hp, L-Am, L-O (within Het), L-S (within Het), R-Hp, R-Am, R-PH, R-O (within Het), R-S (within Het) R-Am, R-Hp, R-PH L-Am, L-Hp L-O, L-S, R-PH, R-O, R-S
24 20/F Bilateral PNH and malformed overlying cortex (6) R-Hp, R-Am, R-PH, R-P (posterior to R-PH toward the Het), R-S and R-O (inserted from O aiming obliquely to Het) Not in MTL contacts R-PH R-PH, R-P, R-S and R-O
Open in a new tab

ACi, anterior cingulate; Am, amygdala/uncus; ANod, anterior nodule; CP, centroparietal; Fe, female; F, frontal lobe; GRe, gyrus rectus; Het, heterotopia; HHE, hemiconvulsion–hemiplegia–epilepsy syndrome; Hp, anterior hippocampus; I, insula; IC, infracalcarine; L, left; M, male; Me, mesial; MCi, mid cingulate gyrus; MRI, magnetic resonance imaging; MTL, mesial temporal lobe; NC, neocortical; O, occipital lobe; OF, orbitofrontal; OP, operculum; P, parietal lobe; PH, middle hippocampus/parahippocampal gyrus; PNH, periventricular nodular heterotopia; PNod, posterior nodule; PO, parietal occipital; Pt, patient; R, right; S (trigone and supramarginal gyrus); SC, supra-calcarine; SEEG, stereo EEG; SMA, supplementary motor area; SOZ, seizure onset zone; sSOZ, secondary seizure onset zone; T, temporal lobe; TP, temporoparietal; TPo, temporal pole; PO, parietooccipital.

Bold print designates MTL contacts.

Table 2 shows the distribution of the three different patterns (C/SC, IRR, and SP) during wakefulness and sleep: Results indicate that the patterns are largely independent of the state (Wilcoxon signed-rank test; p = 0.58). Indeed, most (73 of 96) are the same in the two states. Only in two cases the patterns switch from SP to C/SC (SP during wake and C/SC during sleep in one case and the reverse in a second). In all other cases where differences were observed, the discrepancies are between the “adjacent” patterns, that is, C/SC and IRR or IRR and SP. It can be concluded that the type of background is not strongly affected by wakefulness or sleep. As expected, spike, ripple, and FR median rates are significantly higher during sleep than wakefulness (Wilcoxon signed-rank test, spike, p < 0.001; ripple, p < 0.001; FR, p < 0.001). In what follows the relation between background patterns and the different variables will be reported only for sleep, since wakefulness and sleep show very similar patterns and since most analysis of HFOs is performed during sleep.

Table 2.

Distribution of the different background patterns according to wakefulness and sleep

Wake
C/SC – 29 ch IRR – 34 ch SP – 33 ch
Sleep C/SC – 34 ch 25 8 1
IRR – 27 ch 3 20 4
SP – 35 ch 1 6 28
Open in a new tab

Visual classification of 96 BKG channels during slow-wave sleep (rows) and wake epochs (columns). Evaluation of the intrachannel stability of visual classification of BKG patterns between the different conditions (concordant channels, 73, bold character and nonconcordant channels, 23 samples; Wilcoxon signed-rank test; p = 0.58). ch, channels.

Figure 2 presents examples of the different patterns during sleep. The rhythmicity of the C/SC pattern is clear in the filtered EEG as well as in the corresponding time-frequency maps in which the power is restricted to a narrow frequency range (Fig. 2A). In contrast, for the SP pattern, the EEG is irregular and the corresponding time-frequency maps show energy in all frequency bands (Fig. 2C). The lack of rhythmic activity in the SP pattern does not depend on the gain used to visualize the EEG (Fig. 2D).

Figure 4 shows a histogram of mean wavelet entropy (WE) per channel during sleep. It illustrates that the visual separation in the three categories is reasonably well reflected in the value of WE, which measures the degree of rhythmicity of the signal independently of its amplitude. A low WE value indicates an oscillatory behavior, validating that C/SC is the most oscillatory type of background, whereas SP is the most random. The WE mean SD (median) were 0.88 0.09 (0.86) for C/SC, 0.96 0.09 (0.97) for IRR, and 1.12 0.13 (1.13) for SP patterns. Therefore, the wavelet entropy confirms the visual separation into different patterns.

Figure 4.

Figure 4

Open in a new tab

Histogram of wavelet entropy values for the channels visually marked as belonging to the continuous/semicontinuous category (blue), irregular (green), and sporadic (red). The histogram was obtained by averaging the wavelet entropy of all segments marked as background for each channel. Although there is considerable overlap, the wavelet entropy reflects to a large degree the visual separation into the three patterns.

Epilepsia © ILAE

Background patterns and channel localization. A chi-square test indicated that the type of background pattern was significantly different for the different anatomic locations (χ2(4, N = 96) = 32.7, p ≪ 0.001). When analyzing each type of pattern independently, they were significantly different for the different locations (C/SC: χ2(2, N = 34) = 17.9, p ≪ 0.001; IRR: χ2(2, N = 27) = 7.1, p = 0.03; SP χ2(2, N = 35) = 7.7, p = 0.02). The C/SC pattern was more common in the Hp than in the other structures, whereas the IRR and SP patterns were more common in the amygdala than in the other structures (Fig. 5A).

Figure 5.

Figure 5

Open in a new tab

Background pattern distribution during sleep according to (A) the anatomic localization of the electrodes, (B) the seizure-onset zone, and (C) the presence of a lesion. The C/SC pattern was the most common in the hippocampus (marked with *) and rare in the amygdala. The type of background pattern observed in a channel is not influenced by that channel belonging to the seizure-onset zone or to a lesion.

Epilepsia © ILAE

Background patterns and SOZ

Forty-nine channels belonged to the SOZ, 20 to the sSOZ, and 27 were outside any SOZ (nSOZ). A chi-square test indicated that we cannot reject the hypothesis that the type of background pattern is unaffected by whether the channel belongs to the SOZ (χ2(4, N = 96) = 2.53, p = 0.64) (Fig. 5B).

Background pattern and lesions

Nineteen channels were lesional and 77 were nonlesional. A chi-square test indicated that we cannot reject the hypothesis that the type of background pattern is similar regardless of whether the channel corresponds to a lesion (χ2(2, N = 96) = 1.19, p = 0.55) (Fig. 5C). Therefore, the types of pattern are not influenced by whether a channel is within the SOZ, or whether it is a lesional channel.

Background pattern and spikes

Spike rates were significantly different for the different background patterns (KW test, p = 0.02). Post hoc analysis showed significant difference only between IRR and SP channels. Median rates were highest in channels with the C/SC pattern, followed by channels with the IRR and by the SP pattern (C/SC: 19.4; IRR: 18.4; SP: 11.6). It must be noted that the background sections analyzed did not include any spikes. A negative linear correlation was found between the averaged WE and spike rates for each channel (ρ: 0.53, p ≪ 0.001; Fig. 6A), confirming that spikes are more frequent with the C/SC pattern than with the other patterns.

Figure 6.

Figure 6

Open in a new tab

Relationship between averaged wavelet entropy (WE) of the background of each channel and the rates of spikes, ripples, and fast ripples. Lower WE reflects a more rhythmic background, that is. a pattern like the C/SC pattern. Events are generally more frequent with lower WE.

Epilepsia © ILAE

Background pattern and HFOs

Ripple and FR rates were highest in channels with the C/SC pattern followed by channels with the IRR pattern and then the SP pattern. For ripples, a KW test showed a significant effect ( χ22=67.4, p ≪ 0.001;). Post hoc analysis showed significant differences between all groups (median values: C/SC: 167.3; IRR: 41.1; SP: 5.4). For FR the KW test showed a significant effect ( χ22=34.8, p ≪ 0.001). Post hoc analysis showed significant differences between C/SC and SP and between IRR and SP, but not between C/SC and IRR patterns (median values: C/SC: 20.05; IRR: 7.5; SP: 0.2).

A negative linear correlation was found between the averaged WE and ripple rates for each channel (ρ: 0.78, p ≪ 0.001; Fig. 6B) and between the averaged WE and the FR rates for each channel (ρ: 0.4, p ≪ 0.001; Fig. 6C). There was also a negative correlation between wavelet entropy and spike rate (ρ: 0.53, p ≪ 0.001, Fig. 6A). These negative correlations indicate that the more rhythmic the background pattern, the more likely it is that these events (ripples, fast ripples, spikes) are frequent.

Discussion

Transient interictal HFOs, ripples (80–250 Hz) and FRs (250–500 Hz) are considered reliable markers of epileptogenic tissue (Bragin et al., 2010; Gotman, 2010; Jacobs et al., 2010). Studies have demonstrated the correlation of higher HFOs rates with the SOZ (Jacobs et al., 2008, 2009; Crepon et al., 2010) and with the epileptogenic zone (Jacobs et al., 2010), suggesting their promising role as an effective biomarker for clinical use (Bragin et al., 2010). Ripples and FRs are defined as brief transient oscillatory events that stand out from the background. In several studies they are detected automatically and are required to be several times larger than the background (Staba et al., 2002; Gardner et al., 2007; Crepon et al., 2010; Zelmann et al., 2010). In our studies, we mark HFOs by visual analysis. During the process of this marking, we noticed that some channels presented a continuous or semicontinuous rhythmic activity in the ripple/FR frequency range, particularly in mesial temporal structures. We present here results of a first evaluation of this pattern, clearly establishing its existence (34/96 channels in sleep; 29/96 in wake) and relating it to some anatomic and neurophysiologic variables. We have not assessed the extent to which this pattern is present in the brain, since we have studied only mesial temporal channels.

As HFOs are more frequent during sleep (Staba et al., 2004; Clemens et al., 2007; Bagshaw et al., 2009) one might think that the continuous rhythmic activity could result from a very high rate of HFOs, which merge into each other to make an almost continuous rhythmic activity. We established, however, that the distribution of patterns, from continuous/semicontinuous to irregular and sporadic, was little affected by sleep: the pattern is most often unchanging between wakefulness and sleep and appears, therefore, an intrinsic characteristic of a channel. We have not evaluated whether this pattern is responsive to other physiologic or behavioral changes. We also established that the C/SC pattern is more likely to occur in the anterior hippocampus than in the posterior hippocampus and particularly than in the amygdala, where it is very rare.

There was a tendency for the channels with the C/SC pattern to have a higher spiking rate than the channels with the IRR or SP pattern. The background pattern was specifically assessed in spike-free intervals and this finding cannot therefore be the result of oscillations occurring during spikes. We also excluded 1 s before and after each spike, as high frequency activity is attenuated immediately following spikes (Urrestarazu et al., 2006). Given that we did not find a significant association between the C/SC pattern and the SOZ, we cannot conclude that this pattern represents more epileptogenic tissue than the pattern with sporadic HFOs.

The existence of the continuous or semicontinuous rhythmic pattern raises the question of the existence of HFOs defined as distinct oscillations. In the channels where the C/SC pattern is present, the traditional definition of HFOs reflects most often an event having a frequency similar to that of the background, but an abrupt increase in amplitude (Fig. 3), although we have not evaluated systematically the respective frequency contents of the background in comparison to the distinct HFOs. This continuously oscillating patterns has not been described in experimental animals but it may well exist and has not been observed as a result of the usual definition of HFOs (relative to the background).

For the clinical use of HFOs as biomarkers of epileptogenicity, automatic detectors are necessary. The detection of HFOs in channels with continuous high frequency activity is particularly challenging. The performance of automatic detectors of HFOs may improve when first detecting nonoscillatory baseline segments (using an algorithm based on wavelet entropy, which measures the intrinsic oscillatory nature of the signal independently of amplitude) and then using only these baseline segments to compute an energy threshold to find HFOs. However, the automatic detection of HFOs in channels with C/SC background cannot rely on the detection of nonoscillatory baseline segments. A possibility to detect HFOs in those channels is an iterative approach in which large HFOs are first detected and removed from the signal and smaller HFOs are detected in subsequent iterations (Zelmann et al., 2012).

The findings of this study indicate that the high-frequency content of the intracerebral EEG needs to be fully evaluated, in mesial temporal structures and in the rest of the brain, in the form of distinct brief oscillations and as “background.”

Acknowledgments

This work was supported by grant MOP-102710 of the Canadian Institutes of Health Research.

Footnotes

Disclosure

None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

References

  1. Axmacher N, Elger CE, Fell J. Ripples in the medial temporal lobe are relevant for human memory consolidation. Brain. 2008;131:1806–1817. doi: 10.1093/brain/awn103. [DOI] [PubMed] [Google Scholar]
  2. Bagshaw AP, Jacobs J, LeVan P, Dubeau F, Gotman J. Effect of sleep stage on interictal high-frequency oscillations recorded from depth macroelectrodes in patients with focal epilepsy. Epilepsia. 2009;50:617–628. doi: 10.1111/j.1528-1167.2008.01784.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bragin A, Engel J, Jr, Wilson CL, Fried I, Buzsaki G. High-frequency oscillations in human brain. Hippocampus. 1999a;9:137–142. doi: 10.1002/(SICI)1098-1063(1999)9:2<137::AID-HIPO5>3.0.CO;2-0. [DOI] [PubMed] [Google Scholar]
  4. 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. 1999b;40:127–137. doi: 10.1111/j.1528-1157.1999.tb02065.x. [DOI] [PubMed] [Google Scholar]
  5. Bragin A, Wilson CL, Staba RJ, Reddick M, Fried I, Engel J., Jr Interictal high-frequency oscillations (80–500 Hz) in the human epileptic brain: entorhinal cortex. Ann Neurol. 2002;52:407–415. doi: 10.1002/ana.10291. [DOI] [PubMed] [Google Scholar]
  6. Bragin A, Engel J, Jr, Staba RJ. High-frequency oscillations in epileptic brain. Curr Opin Neurol. 2010;23:151–156. doi: 10.1097/WCO.0b013e3283373ac8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chander R. MEng thesis. McGill University; 2007. Algorithms to detect high frequency oscillations in human intracerebral EEG; p. 123. [Google Scholar]
  8. Clemens Z, Molle M, Eross L, Barsi P, Halasz P, Born J. Temporal coupling of parahippocampal ripples, sleep spindles and slow oscillations in humans. Brain. 2007;130:2868–2878. doi: 10.1093/brain/awm146. [DOI] [PubMed] [Google Scholar]
  9. Crepon B, Navarro V, Hasboun D, Clemenceau S, Martinerie J, Baulac M, Adam C, Le Van Quyen M. Mapping interictal oscillations greater than 200 Hz recorded with intracranial macroelectrodes in human epilepsy. Brain. 2010;133:33–45. doi: 10.1093/brain/awp277. [DOI] [PubMed] [Google Scholar]
  10. Gardner AB, Worrell GA, Marsh E, Dlugos D, Litt B. Human and automated detection of high-frequency oscillations in clinical intracranial EEG recordings. Clin Neurophysiol. 2007;118:1134–1143. doi: 10.1016/j.clinph.2006.12.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gotman J. High frequency oscillations: the new EEG frontier? Epilepsia. 2010;51(Suppl 1):63–65. doi: 10.1111/j.1528-1167.2009.02449.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Jacobs J, LeVan P, Chander R, Hall J, Dubeau F, Gotman J. Interictal high-frequency oscillations (80–500 Hz) are an indicator of seizure onset areas independent of spikes in the human epileptic brain. Epilepsia. 2008;49:1893–1907. doi: 10.1111/j.1528-1167.2008.01656.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Jacobs J, Levan P, Chatillon CE, Olivier A, Dubeau F, Gotman J. High frequency oscillations in intracranial EEGs mark epileptogenicity rather than lesion type. Brain. 2009;132:1022–1037. doi: 10.1093/brain/awn351. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Jacobs J, Zijlmans M, Zelmann R, Chatillon CE, Hall J, Olivier A, Dubeau F, Gotman J. High-frequency electroencephalographic oscillations correlate with outcome of epilepsy surgery. Ann Neurol. 2010;67:209–220. doi: 10.1002/ana.21847. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Jacobs J, Kobayashi K, Gotman J. High-frequency changes during interictal spikes detected by time-frequency analysis. Clin Neurophysiol. 2011;122:32–42. doi: 10.1016/j.clinph.2010.05.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Kobayashi K, Jacobs J, Gotman J. Detection of changes of high-frequency activity by statistical time-frequency analysis in epileptic spikes. Clin Neurophysiol. 2009;120:1070–1077. doi: 10.1016/j.clinph.2009.03.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Nagasawa T, JuhQsz C, Rothermel R, Hoechstetter K, Sood S, Asano E. Spontaneous and visually-driven high-frequency oscillations in the occipital cortex: intracranial recording in epileptic patients. Hum Brain Mapp. 2012;33:569–583. doi: 10.1002/hbm.21233. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Rosso OA, Blanco S, Yordanova J, Kolev V, Figliola A, Schurmann M, Basar E. Wavelet entropy: a new tool for analysis of short duration brain electrical signals. J Neurosci Methods. 2001;105:65–75. doi: 10.1016/s0165-0270(00)00356-3. [DOI] [PubMed] [Google Scholar]
  19. Staba RJ, Wilson CL, Bragin A, Fried I, Engel J., Jr Quantitative analysis of high-frequency oscillations (80–500 Hz) recorded in human epileptic hippocampus and entorhinal cortex. J Neurophysiol. 2002;88:1743–1752. doi: 10.1152/jn.2002.88.4.1743. [DOI] [PubMed] [Google Scholar]
  20. Staba RJ, Wilson CL, Bragin A, Jhung D, Fried I, Engel J., Jr High-frequency oscillations recorded in human medial temporal lobe during sleep. Ann Neurol. 2004;56:108–115. doi: 10.1002/ana.20164. [DOI] [PubMed] [Google Scholar]
  21. Urrestarazu E, Jirsch JD, LeVan P, Hall J, Avoli M, Dubeau F, Gotman J. High-frequency intracerebral EEG activity (100–500 Hz) following interictal spikes. Epilepsia. 2006;47:1465–1476. doi: 10.1111/j.1528-1167.2006.00618.x. [DOI] [PubMed] [Google Scholar]
  22. Urrestarazu E, Chander R, Dubeau F, Gotman J. Interictal high-frequency oscillations (100–500 Hz) in the intracerebral EEG of epileptic patients. Brain. 2007;130:2354–2366. doi: 10.1093/brain/awm149. [DOI] [PubMed] [Google Scholar]
  23. Worrell GA, Gardner AB, Stead SM, Hu S, Goerss S, Cascino GJ, Meyer FB, Marsh R, Litt B. High-frequency oscillations in human temporal lobe: simultaneous microwire and clinical macroelectrode recordings. Brain. 2008;131:928–937. doi: 10.1093/brain/awn006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Wu JY, Sankar R, Lerner JT, Matsumoto JH, Vinters HV, Mathern GW. Removing interictal fast ripples on electrocorticography linked with seizure freedom in children. Neurology. 2010;75:1686–1694. doi: 10.1212/WNL.0b013e3181fc27d0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Zelmann R, Mari F, Jacobs J, Zijlmans M, Chander R, Gotman J. Automatic detector of high frequency oscillations for human recordings with macroelectrodes. Conf Proc IEEE Eng Med Biol Soc. 2010;2010:2329–2333. doi: 10.1109/IEMBS.2010.5627464. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Zelmann R, Mari F, Jacobs J, Zijlmans M, Dubeau F, Gotman J. A comparison between detectors of high frequency oscillations. Clin Neurophysiol. 2012;123:106–116. doi: 10.1016/j.clinph.2011.06.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Zijlmans M, Jacobs J, Zelmann R, Dubeau F, Gotman J. High-frequency oscillations mirror disease activity in patients with epilepsy. Neurology. 2009;72:979–986. doi: 10.1212/01.wnl.0000344402.20334.81. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES