Epilepsy & Behavior 15 (2009) 22–33

Contents lists available at ScienceDirect

Epilepsy & Behavior

journal homepage: www.elsevier.com/locate/yebeh

Review

The role of EEG in epilepsy: A critical review

Soheyl Noachtar *, Jan Rémi
Epilepsy Center, Department of Neurology, University of Munich, Marchioninistrasse 15, 81377 Munich, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Electroencephalography (EEG) is the most specific method to define epileptogenic cortex. Its sensitivity
Received 18 February 2009 and specificity depend on several factors such as age and recording procedures, for example, sleep record-
Accepted 19 February 2009 ings and activation procedures (hyperventilation, photic stimulation). EEG reveals characteristic findings
Available online 25 February 2009
in several epilepsy syndromes. Rarely, epileptiform discharges are recorded in healthy, particularly young
individuals. Ictal video/EEG recording is considered to be critical in localizing the epileptogenic zone. A
Keywords: careful analysis of the first clinical signs and symptoms of a seizure and of the evolution of the seizure
EEG
symptomatology can provide important localizing clues. Although surface EEG recordings are less sensi-
Epilepsy
Epileptiform discharges
tive than invasive studies, they provide the best overview and, therefore, the most efficient way to define
Invasive EEG the approximate localization of the epileptogenic zone. Invasive recordings are used in patients in whom
EEG-video-monitoring the epileptogenic zone either cannot be located with noninvasive diagnostic methods or is adjacent to
eloquent cortex. The most commonly used invasive electrodes are stereotactically implanted depth elec-
trodes and subdural strip or grid electrodes. Foramen ovale and epidural electrodes are of intermediate
invasiveness, but less sensitive. Invasive electrodes are subject to sampling errors if misplaced and should
be used only after exhaustive noninvasive evaluations have (1) failed to localize the epileptogenic zone
and (2) led to a testable hypothesis regarding this localization. Invasive EEG studies are associated with
additional risks that are justifiable only if there is a good chance of obtaining essential localizing infor-
mation and on a potentially resectable area.
Ó 2009 Elsevier Inc. All rights reserved.

1. Introduction
Electroencephalography (EEG) is the most useful diagnostic
procedure for epilepsy, and with the advent of more sophisticated
methods of imaging structural damage, epilepsy is now actually
one of the few common clinical problems routinely demanding
EEG evaluation [1]. EEG can answer three main questions in the
diagnostic workup of patients suspected of having epilepsy [1]:
Does the patient have epilepsy?
Where is the epileptogenic zone?
How good is therapy?
2. Does the patient have epilepsy?
Interictal epileptiform discharges (IEDs) in EEG help to differen-
tiate between epileptic and other nonepileptic paroxysmal attacks.
The following EEG patterns are considered epileptiform [2,3]:
* Corresponding author. Fax: +49 89 7095 6691.
E-mail address: noa@med.uni-muenchen.de (S. Noachtar).
Spikes
Sharp waves
Benign epileptiform discharges of childhood
Spike–wave complexes
Slow spike–wave complexes
3-Hz spike–wave complexes
Polyspikes
Hypsarrhythmia
Seizure pattern
Status pattern
There is no difference in the diagnostic information between
sharp waves and spikes [4] (Fig. 1). Rarely, the electroencephalo-
gram of a normal individual shows IEDs. No epileptiform dis-
charges were seen in a study with 100 healthy volunteers [5]. A
study in more than 13.000 young male candidates for aircrew per-
sonnel showed IEDs in 0.5% (in 58% of these only during photic
stimulation) [6]. Forty-three of these individuals were followed
up to 29 years and only one person developed epilepsy. The
authors conclude that the chance of developing epilepsy in healthy
individuals with IEDs on EEG is 2–3% [6]. IEDs are more common in
normal children between 1 and 15 years of age (about 1–2%) [7].
Mostly they are so-called benign focal epileptiform discharges of
childhood and may occur in the occipital, centrotemporal, and

1525-5050/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.yebeh.2009.02.035
 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33 23

Fig. 1. Left temporal sharp wave with a negative sharp component at electrodes F7 and T7 in a bipolar longitudinal recording during light sleep. This 32-year-old patient had
left temporal lobe epilepsy due to mesial left temporal sclerosis.

Fig. 2. Right central benign epileptiform discharges of childhood in a bipolar longitudinal recording. This 8-year-old child had rare generalized tonic–clonic seizures during
sleep that were preceded by left face clonic seizures.

frontal regions [8] (Fig. 2). Younger children more often display
occipital discharges. Only about 8% of these children develop epi-
lepsy [9]. About 2% of patients with neurological disorders but no
epilepsy have IEDs [10]. Depending on age and epileptogenicity,
epileptiform discharges occur in up to 98% of patients with epi-
lepsy [11].
Initially, EEG may be normal in 12–50% of patients with epi-
lepsy [12,13]. Repeated EEG, however, increases the diagnostic
yield [14]. Long-term recordings likewise increase the chance of
recording IEDs. Activation procedures such as hyperventilation,
photic stimulation, and sleep deprivation enhance the diagnostic
sensitivity of EEG. EEG including wake and sleep following partial
sleep deprivation showed IEDs in about 50% of patients with a
clinically likely diagnosis of epilepsy, in whom routine EEG
including stage II sleep did not reveal any IEDs [15]. However,
the value of routine EEG after a first unprovoked seizure has been
debated. Some authors would not start therapy following a first
seizure [16]. Others emphasize that the risk of seizure relapse is
increased if EEG shows IEDs [17]. The decision for antiepileptic
therapy clearly depends on the etiology of the epilepsy and the
individual seizure relapse risk, rather than on a general rule to
start therapy after a first or second seizure.
24 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33

There is no objective definition of IEDs [3]. Even experienced Table 2

electroencephalographers sometimes disagree on the diagnosis of Typical interictal epileptiform discharges on the EEGs of patients with characteristic
epilepsy syndromes or etiologies.
IEDs, and, therefore, EEG interpretation is hampered by poor inter-
observer reliability [18]. Normal sharp transients have to be differ- EEG pattern Epileptic syndrome/etiology
entiated from epileptiform discharges to avoid misinterpretation of Anterior temporal spikes Mesial temporal lobe
the EEG recording. Some EEG patterns that are not epileptic may be epilepsy
overinterpreted, leading to a misdiagnosis of epilepsy [1]. Table 1 Generalized 3-Hz spike–wave complexes Absence epilepsy
>4-Hz spike–wave complexes, generalized Juvenile myoclonic epilepsy
summarizes several sharp variants (wickets, spikes, 14- and 6-Hz polyspikes
positive spikes, etc.) and their characteristics which can be easily Generalized slow spike–wave complexes Lennox–Gastaut syndrome
confused with epileptiform discharges (Fig. 3). The polarity of Regional (extratemporal) polyspikes Focal cortical dysplasia
sharp transients is important for interpretation; the sharp compo- Hypsarrhythmia West syndrome
nent of epileptiform discharges is typically negative (Fig. 1) [3]. Po-
sitive spikes are rare and can be found in patients who have
undergone resective brain surgery [19]. to differentiating between ictal and interictal EEG findings. How-
ever, although this distinction is clinically important, it is very
3. Where is the epileptogenic zone? arbitrary. It is easy if, for instance, a clonic seizure is observed dur-
ing an EEG discharge, but it may be difficult to assess the level of
Several characteristic EEG patterns are associated with well-de- consciousness during an epileptiform discharge [20]. A continuum
fined epilepsy syndromes (Table 2). EEG, thus, helps to define cer- between the interictal and ictal states for both focal and general-
tain epilepsy syndromes, which is important for selection of ized epilepsies is clinically more likely than a clear-cut border line
therapy and assessment of prognosis of the epilepsy. We are used (Fig. 4). A reaction time task during EEG (klicker test) helps to bet-

Table 1
Sharp transients that represent normal EEG variants and can be easily confused with epileptiform discharges.

Rhythmical temporal Wicket spike Small sharp spike 14- and 6-Hz 6-Hz ‘‘spike and wave” SREDA
theta of drowsiness positive ‘‘spikes”
Frequency (Hz) 4–7 6–12 Sporadic (about 14 and 6 5–7 5–6
50 ms)
Localization Temporal Temporal Frontal maximum Lateral to posterior Generalized Generalized
temporal
Waveform Notched, rhythmic Monophasic,similar Amplitude <50 lV Monophasic Diphasic, small spike Sudden onset and
to l waves duration <50 ms and large wave sudden end
Level of Relaxed wake sleep Wake sleep stage 1 Relaxed wake, sleep Wake sleep stages 1 Sleep stage 1 Wake sleep stage 1/
consciousness stage 1 stages 1 and 2 and 2 hyperventilation
Age Young adults Adults Adults Adolescents, adults Adolescents, adults Elderly
Duration 10 s 0.5–2 s Single discharges <1 s <1 s 40–80 s
Characteristics Synonym: Modulated, similar to Synonym: benign 6- and 14-Hz ‘‘Phantom” spike and Subclinical rhythmic
psychomotor variant sleep spindles epileptiform components may wave. Occipital or electrographic
transients of sleep occur separately frontal maximum discharge of adults
(‘‘BETS”)

Fig. 3. Wicket spikes (arrow) in the left temporal region in a bipolar longitudinal recording during wakefulness. This patient had migraine without aura.
 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33 25

lamotrigine, topiramate), whereas other AEDs may lead to deterio-

Status ration of the condition (carbamazepine, tiagabine) (Fig. 5). Several
EEG seizure pattern
epilepticus generalized epilepsy syndromes are associated with typical epilep-
tiform discharges (Table 2). Generalized SWCs with a repetition
rate of 3 Hz are typically associated with clinical absence seizures
Clinical symptoms

Ictal EEG seizure pattern if the discharge lasts longer than 4 seconds (Fig. 5) [24]. However,
selected patients may still be able to respond, though more slowly
than normal (Fig. 6). Other patients may already not respond to
auditory stimuli during SWCs of 1–2 seconds’ duration (Fig. 7). A
Subclinical EEG seizure pattern standardized reaction test to auditory or visual stimuli is manda-
tory to assess responsiveness during IEDs and is frequently re-
quired to assess the driving capacity of patients with absence
epilepsy [21].
Interictal Interictal epileptiform discharges
Generalized polyspikes are common in patients with juvenile
myoclonic epilepsy (JME). These patients are more likely to be pho-
time tosensitive (30%) than patients with other generalized idiopathic
epilepsies such as absence epilepsies [25]. About 30–60% of pa-
Fig. 4. Continuum between the interictal state and ictal state in patients with
epilepsy. tients with idiopathic generalized epilepsies display regional IEDs
and slowing in some EEG recordings on long-time follow-up
(Fig. 7) [26,27]. Thus, regional IEDs in the presence of generalized
ter assess responsiveness during epileptiform discharges (Fig. 5) IEDs do not exclude the diagnosis of idiopathic generalized epi-
[21]. This information is clinically important, for instance, to eval- lepsy. Sleep modulates IEDs in these patients as more irregular
uate a patient’s ability to drive a car. SWCs and polyspikes were recorded than during waking periods.
Sometimes EEG seizure patterns are recorded in clinically Generalized slow spike–wave complexes (SSWCs) are typical of
asymptomatic patients. These are called subclinical seizures and patients with Lennox–Gastaut syndrome, which has a poor prog-
are probably the case when the epileptic seizure discharge involves nosis (Fig. 8) [28,29]. EEG during sleep is more likely to show gen-
only asymptomatic cortex or when the clinical testing does not eralized SSWCs than EEG during waking, and typically, generalized
cover the cortical function that is altered [22] (Fig. 4). Spread of polyspikes also occur during sleep in these patients. Some of these
the epileptic activity into symptomatogenic cortex eventually patients had West syndrome at an earlier age. The typical EEG
leads to clinical symptoms. recording of West syndrome is hypsarrhythmia.

3.1. Generalized epilepsies 3.2. Focal epilepsies

Generalized 3-Hz spike–wave complexes (SWCs) are the hall- The precise localization of the epileptiform discharges is crucial
mark of absence epilepsy [23], which has a very good prognosis to the localization of the epileptogenic zone, particularly in pa-
and responds well to certain antiepileptic drugs (AEDs: valproate, tients considered for resective epilepsy surgery [30]. In the early

Fig. 5. Generalized spike–wave complexes in a 25-year-old patient with childhood absence epilepsy who had a seizure relapse after being seizure free off medication for
8 years. This bipolar longitudinal recording demonstrates that the patient is able to respond to an auditory stimulus to push a button at the onset of the discharge, but fails to
respond to later stimuli. At the end of the discharge a delayed response was recorded.
26 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33

Fig. 6. Generalized spike–wave complexes in an ipsilateral ear reference recording during wakefulness in a 19-year-old patient with juvenile absence epilepsy. Note the
higher amplitude in the left frontal region at onset of the discharge. Generalized spike–wave complexes typically have a shifting maximum between left and right frontal
regions. Despite the generalized spike–wave complexes, the patient was initially able to respond to the auditory stimuli. The response time, however, gradually increased
with the duration of the discharge. At the end of the discharge, the patient failed to respond to the last stimulus.

Fig. 7. Generalized polyspikes and spike waves in a 39-year-old patient with juvenile absence epilepsy who never was free of absence seizures. Generalized tonic–clonic
seizures ceased about 20 years ago. This bipolar longitudinal recording shows generalized polyspikes and spike waves in the left part of the figure. The patient fails to respond
to auditory stimuli during the discharge (klicker), but he is able to do so after the discharge. In the right part of the figure, left temporal spike waves were recorded (arrow).

days of EEG, it was recognized that anterior temporal IEDs were the etiology of epilepsy in many patients and to consider surgical
frequently associated with temporal lobe epilepsy (TLE) [31]. The treatment, as the majority of these patients are not controlled with
advent of MRI enabled us to identify mesial temporal sclerosis as medical treatment only [32]. The use of anterior temporal and
 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33
Fig. 8. Generalized slow spike–wave complexes in a 22-year-old patient with idiopathic Lennox–Gastaut syndrome. Note the slow repetition rate (<2.5 Hz) of the spike–wave
complexes in an ipsilateral ear reference montage.
sphenoidal electrodes increases the yield of EEG in patients with
mesial temporal epilepsy [33,34]. Patients with TLE frequently
have IEDs independently in both temporal lobes. The occurrence
of strictly unilateral temporal IEDs has an excellent predictive va-
lue for successful resective epilepsy surgery in TLE [35]. The pre-
dominance of IEDs in the mesial or lateral temporal regions
points to a mesial or lateral TLE as evidenced by MRI lesion or inva-
sive EEG [36]. Patients with mesial TLE as a result of hippocampal
sclerosis who show frequent IEDs have a poorer prognosis for
resective epilepsy surgery (28.6% seizure free) than patients with
rare IEDs (80.5% seizure free) [37].
Mesial temporal spikes do not, however, necessarily indicate
TLE. EEGs in patients with extratemporal epilepsies also have a
tendency to exhibit temporal IEDs. Despite this tendency, typically
the IEDs are located predominately in the epileptogenic lobe. How-
ever, some patients with extratemporal epilepsies do not show any
IEDs [38]. EEGS in patients with central (46.7%) and frontal (17.8%)
lesions more often lack interictal spiking than patients with tem-
poral (4.2%) or parieto-occipital (7.0%) lesions (p < 0.01). IEDs were
found exclusively in the lesioned lobe in 45.6% of temporal, 31.3%
of central, 25% of frontal, and only 5% of parieto-occipital lesions
(p < 0.01). Ictal EEG changes in the lesioned lobe occurred more
commonly in temporal (58%), frontal (43.8%), and central (42.1%)
than in parieto-occipital (28.9%) lesions (p < 0.05). The occurrence
and localization of ictal and interictal epileptiform EEG discharges
in patients with clearly defined lesions on MRI differ depending on
the lesioned lobe. In patients with temporal lesions, the EEG
changes are most congruent with the location of the lesion, and
parieto-occipital lesions have the least congruent changes on
EEG. Regional polyspikes seem to provide a useful etiological as-
pect mainly for extratemporal epilepsy, as they occur significantly
more commonly in patients with cortical malformations than in
those with other etiologies [39].
Ictal EEG recordings have the highest diagnostic yield, particu-
larly if accompanied by simultaneous video recordings of the pa-
tient. Therefore, most centers include ictal recordings in their
diagnostic evaluation of patients considered for epilepsy surgery
[40–42]. The localizing yield of ictal EEG is higher in TLE than in
27
extratemporal epilepsies [43]. Surface ictal EEG adequately local-
ized onsets in 72% of cases, more often in temporal than extratem-
poral epilepsy. Lateralized ictal onsets were seen in 57% of seizures
and were most common in mesial temporal lobe epilepsy (MTLE),
lateral frontal lobe epilepsy (LFLE), and parietal lobe epilepsy,
whereas lateralized onsets predominated in neocortical temporal
lobe epilepsy and generalized onsets in mesial frontal lobe epilepsy
(MFLE) and occipital lobe epilepsy [43]. Approximately two-thirds
of seizures were localized, 22% generalized, 4% lateralized, and 6%
mislocalized/lateralized. False localization/lateralization occurred
in 28% of occipital and 16% of parietal seizures. Rhythmic temporal
theta at ictal onset was seen exclusively in temporal lobe seizures,
whereas localized repetitive epileptiform activity was highly pre-
dictive of LFLE. Seizures arising from the lateral convexity and me-
sial regions were differentiated by a high incidence of repetitive
epileptiform activity at ictal onset in the former and rhythmic the-
ta activity in the latter. With the exception of MFLE, ictal record-
ings are very useful in the localization/lateralization of focal
seizures. One limitation of ictal EEG is the potential for false local-
ization/lateralization in occipital and parietal lobe epilepsies [43].
Regional slowing of EEG activity is not specific for epilepsy and,
therefore, does not support the diagnosis of epilepsy. However,
rhythmic slowing may assist in localization of the epileptogenic
zone once the diagnosis of epilepsy is established. Interictal rhyth-
mic slowing in the temporal regions (temporal intermittent rhyth-
mic delta activity, TIRDA) occurs in patients with focal epilepsies
[44]. Rhythmical midline theta (RMT) activity not related to drows-
iness or mental activation is significantly more common in patients
with frontal (48.1%) than with temporal (3.7%) lobe epilepsy [45].
Controls without epilepsy did not have this kind of RMT activity.
This finding is in clear contrast to the initial description of the pat-
tern which was attributed to TLE [46]. However, it was later
pointed out that drowsiness leads to the same EEG pattern which
was not clearly excluded in the early description, and one has to
consider that localization of the epileptogenic zone relied on sei-
zure semiology and interictal EEG only; imaging studies (CT, MRI,
PET, SPECT) or video/EEG monitoring were not available at that
time. The localizing value of RMT activity is even more important
28 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33
in those patients with FLE who do not have any IEDs at all (24%),
which is more frequently the case than in TLE. RMT was observed
in the majority of these patients with FLE (62%) [45]. Interictal RMT
activity is, thus, common and has a localizing value in patients
with FLE, provided that conditions such as drowsiness and mental
activation as confounding factors for RMT activity are excluded
[45].
3.3. Invasive EEG recordings
In most patients who are referred to an epilepsy surgery center
today, noninvasive methods are sufficient to evaluate the patients
before surgical resection [30,47]. This is especially the case when
findings from imaging studies (MRI, PET) are congruent with the
electroclinical and neuropyschological findings and when the epi-
leptogenic zone is in one mesial temporal lobe. Invasive EEG
recordings are used only when noninvasive methods do not allow
sufficient localization of the epileptogenic zone or when the epi-
leptogenic zone is anatomically too close to eloquent cortex
[30,48]. The use of invasive electrodes and with it the risk of com-
plications [49] are justified only when noninvasive methods (MRI,
video/EEG monitoring, FDG-PET, and ictal SPECT) deliver evidence
that the epileptogenic zone is potentially resectable. Invasive EEG-
recording is therefore based on the following criteria:
There exists a clear hypothesis on the epileptogenic zone.
The hypothesis can be tested with the chosen electrodes.
The suspected epileptogenic zone is potentially resectable.
There is a low risk of possible complications.
The typical electrodes are illustrated in Fig. 9.
3.3.1. Advantages of invasive electrodes
Invasive electrodes have several advantages that justify their
use. The signal-to-noise ratio is considerably better with invasive
EEG than with surface EEG. An invasive EEG derivation may al-
ready show a seizure pattern while the surface EEG seems still
undisturbed. The higher sensitivity of foramen ovale electrodes
as compared with surface EEG is illustrated in Fig. 10. Electromyo-
graphic artifacts from muscle activity or ballistic movement arti-
facts can disturb surface EEG in typical patterns. The artifacts are
considerably reduced with ictal invasive EEG recordings (Fig. 11).
In the following, the typical constellations and questions in the
use of invasive EEG electrodes are discussed.
3.3.2. Constellations for the use of invasive EEG electrodes
Invasive electrodes may help to determine the extent of the epi-
leptogenic zone and to distinguish the epileptogenic zone from le-
sions identified with imaging studies and from eloquent cortex.
The epileptologic constellations where EEG electrodes are used
can be summed up as follows:
Unitemporal versus bitemporal
Unilateral temporal versus unilateral extratemporal
Unilateral extratemporal versus bifocal/multifocal
3.3.2.1. Unitemporal versus bitemporal. The most common task
accomplished with invasive electrodes is distinguishing unilateral
from bilateral temporal lobe epilepsy. In one study, for a large per-
centage (77%) of the patients with temporal lobe epilepsies in
whom lateralization was not achieved with noninvasive methods,
seizure patterns from only one temporal lobe were obtained in
invasive EEG studies [50]. The interictal and ictal EEG findings in
that study did not allow for lateralization. Almost a third of these
Subdural
electrodes
Epidural
electrodes
Stereotactically
implanted
depth electrodes
Foramen-ovale
electrodes
Fig. 9. Schematic representation of different invasive electrode types.
patients achieved seizure freedom after a unilateral anterior tem-
poral lobe resection [51]. In another third of these patients, inva-
sive EEG demonstrated an extratemporal origin of the seizure
pattern, which is congruent with more recent studies [52].
Interictal epileptiform discharges contralateral to the lobe
where seizure patterns were recorded reduce the chance for sei-
zure freedom after resection, but are in themselves not an indica-
tion for invasive EEG diagnostics when the other findings (MRI,
ictal EEG, neuropsychological findings, FDG-PET, ictal SPECT) are
consistent with unitemporal epilepsy [35].
3.3.2.2. Unilateral temporal versus extratemporal. Another common
part of presurgical evaluation is determination of the hemisphere
in which the seizure pattern originates. The temporal and frontal
lobes are the most common regions to be evaluated in that respect
[30]. Determining the epileptogenic zone is more difficult when
there is no lesion on MRI on which the location of the invasive elec-
trodes can be based. The development of new imaging techniques
(FDG-PET, flumazenil-PET, ictal SPECT) further limited the use of
invasive EEG recordings [53–55]. Despite some reservations over
the specificity of functional imaging findings, which naturally can
display only the secondary results of epileptic activity [56], and
can be misleading because of spread phenomena, they are helpful,
in our experience, in narrowing down results and finding the right
placement for invasive electrodes. In some cases, the number of
invasive electrodes and, thus, the risks of complications of the
implantation can be reduced.
Fig. 12 is an ictal invasive EEG-recording of a patient with right
frontal focal cortical dysplasia. It demonstrates a focal seizure on-
set at one electrode overlying the cortical dysplasia. The spread of
seizure activity from mesial temporal to ipsilateral frontal regions
is captured with depth electrodes implanted in both lobes (Fig. 13).
The semiology of hypermotor seizures had been very suggestive of
frontal lobe epilepsy [57], but the invasive recording may demon-
strate seizure activity beginning at the amygdala and subsequently
spreading rapidly to the ipsilateral frontal lobe (Fig. 13).
3.3.2.3. Unilateral extratemporal versus bifocal or multifocal. Invasive
diagnostic procedures are more complex when the contralateral
hemisphere also has to be investigated. Typical questions are
whether the epileptogenic zone is located frontally or temporally
on one or on both sides. A given patient is a good candidate for epi-
lepsy surgery only if an epileptogenic zone in one hemisphere
 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33 29

Fig. 10. Synchronous EEG recording with foramen ovale (upper part of the tracing) and scalp electrodes. An EEG seizure pattern can be recorded in electrode Fo8 which is not
reflected in the scalp electrodes.

Fig. 11. Synchronous EEG recording with epidural (upper part of the tracing) and scalp electrodes, which were implanted through single bore holes [59]. While the surface
EEG is obscured from the strong ictal muscle activity of a bilateral tonic seizure, the epidural electrodes demonstrate a left frontopolar (FP1) seizure pattern that spreads to
the adjacent epidural electrodes in a typical fashion.
30 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33

Fig. 12. This selection of a multichannel EEG recording with subdural electrodes in a referential montage to the subdural reference electrode A1 shows a focal seizure onset in
electrode A16, which was overlying the right lateral frontal cortex (arrow, see radiograph). This patient had right frontal lobe epilepsy due to focal cortical dysplasia (see MRI).

Fig. 13. This selection of a multichannel EEG recording with stereotactically implanted depth electrodes in the amygdala (AM), anterior hippocampus (AH), posterior
hippocampus (PH), and orbitofrontal (OF) and mesial frontal (MF) regions demonstrates rapid spread of seizure onset in the amygdala (electrode AM2) to the frontal regions.
The patient had hypermotor seizures that were rarely preceded by abdominal auras. At the time the clinical semiology was a hypermotor seizure; the seizure pattern had
already spread to the left frontal lobe, which could be shown with the depth electrodes. The radiograph demonstrates electrode positioning.

(even when the focus is in more than one lobe) can be identified.
Patients with epileptogenic zones in both hemispheres will most
likely not benefit from a surgical approach. Callosotomy as a palli-
ative approach has been performed in some of these patients.
However, the results from several centers differ considerably
[58,59].
3.3.3. Invasive electrodes
For the most part, the aforementioned invasive electrodes are
either placed subdurally on the cortex (strip or grid electrodes)
or stereotactically implanted into the brain parenchyma. Fig. 9
illustrates the most common electrode types. Epidural and fora-
men ovale electrodes are sometimes referred to as ‘‘semi-invasive”
because their placement is less invasive than that of subdural or
depth electrodes [60–63].
The decision on which electrode type to use depends on the
brain region and the question to be answered, on the one hand,
and the experience and common practice at the different epilepsy
centers, on the other hand. The advantages and disadvantages of
each electrode type are discussed next.
 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33
3.3.3.1. Depth electrodes. Depth electrodes are stereotactically im-
planted, and the technique has considerably improved from angi-
ography and ventriculography toward less invasive and safer
three-dimensional MRI coordinate acquisition. The final placement
of the electrodes can now be determined within the range of a few
millimeters. The placement trajectories are chosen to avoid dam-
age to important brain structures or blood vessels.
Depth electrodes are most commonly used in the temporal lobe.
The electrodes can either be implanted from the lateral face toward
the mesial structures or from occipital positions toward mesial
temporal structures in a sagittal trajectory. The former allows dis-
tinguishing between mesial and lateral temporal activity; the latter
can identify parieto-occipital spread patterns. Basal frontal or me-
sial frontal and parietal areas can be investigated very well with
depth electrodes, whereas the placement of subdural electrodes
can be technically challenging in these locations. The approach
should be guided by the clinical question. Depth electrodes can
also be used in patients where dura and cortex may be attached
after bacterial meningitides, hemorrhages, or trauma.
3.3.3.2. Subdural strip or grid electrodes. Subdural strip electrodes
can be implanted through bore holes to investigate the basal, me-
sial, and lateral temporal cortical surface. They can also be im-
planted in the interhemispheric surface after trepanation.
Subdural grid electrodes also require a craniotomy. Subdural grid
electrodes are especially valuable because large, connected areas
can be examined and exact cortical mapping is possible (Figs. 5
and 8). One of their disadvantages is that the cortex in deep sulci
may not be covered. Only spread out of the sulcus will be visible
in the electrode derivations. Also, after trauma, hemorrhages, or
infections, the dura and cortex may be inseparable and the implan-
tation of subdural electrodes may not be possible.
3.3.3.3. Epidural electrodes. A less invasive method to locate the epi-
leptogenic focus employs epidural electrodes [61]. Epidural elec-
trodes are placed on the dura through bore holes. Typically,
penlike or mushroom-shaped electrodes are used. Some centers
use strips with several electrodes that are epidurally inserted
through a bore hole. The mushroom-shaped electrodes require
one bore hole per electrode. Epidural electrodes were used to cover
larger areas of the convexity, but mesial cortex areas could not be
investigated and basal temporal cortex could be investigated only
with epidural strip electrodes. Their use has steadily declined since
the advent of functional imaging studies (PET, ictal SPECT) and the
improvement of MRI techniques [61]. Their disadvantages are that
they cannot be used for electrical stimulation of the cortex and that
their signal-to-noise ratio is not as favorable compared with that of
subdural electrodes.
3.3.3.4. Foramen ovale electrodes. Foramen ovale electrodes come to
rest in the cisterna ambiens and ideally receive their electrical sig-
nal from the parahippocampal gyrus and neighboring areas. They
were introduced to reduce the risk of intraparenchymal depth elec-
trodes [64]. The implantation closely follows the method used for
blocking the stellate ganglion in patients with trigeminal neuralgia.
Foramen ovale electrodes can record signals from the mesial tem-
poral lobe areas and are an alternative to depth electrodes, which
are implanted in the brain parenchyma [60]. They are used to lat-
eralize temporal lobe epilepsies. In extratemporal epilepsies they
can be used to document the spread to the temporal lobe. Compli-
cations are described to be low, even in patients with previous sur-
gery [65].
3.3.3.5. Disadvantages and risks of invasive electrodes. Even though
surface EEG recordings are less sensitive in detecting epileptiform
31
discharges than invasive EEG recordings, they do give the best
overview and allow estimation of the epileptogenic zone.
Invasive electrodes can only derive a signal in close proximity to
the electrode, typically a few millimeters, so an epileptogenic zone
that is just millimeters further away may be missed. In these cases
the invasive electrodes will show the spread of the seizure pattern
and not the actual origin. When invasive electrodes do not cover
the cortical area where the seizure originates, the electrodes may
show the spread pattern after the clinical seizure begins.
Invasive EEG recordings lead in 1–4% of patients to complica-
tions (hemorrhages or infections with consecutive hemiparesis,
hemianopsia, or aphasia) [49,66], which are mostly temporary,
but should only be accepted as risks when a clear gain in informa-
tion on seizure localization is expected. Invasive electrodes should
therefore be used only after extensive noninvasive EEG studies
have been performed and did not localize the epileptogenic zone
well enough, but they were the grounds on which a testable
hypothesis of the seizure onset zone could be built [30].
3.3.4. Electrocorticography
Intraoperative electrocorticography (ECoG) was introduced sev-
eral decades ago to determine the extent of resection, mostly in
partial temporal lobe resections. The resection could be enlarged
if epileptiform activity could still be recorded from the remaining
tissue. As for intraoperative electrical stimulation, ECoG has strict
time constraints and is usually limited to less than 30 minutes of
recording. The importance of resecting areas with interictal epilep-
tiform activity is not well studied, and the zone from which inter-
ictal epileptiform activity can be recorded is typically far larger
than the actual seizure onset zone. ECoG most likely does not con-
tribute much to standardized procedures like an anterior temporal
lobe resection, but may be helpful in determining the extent of
resection in more diffuse pathologies such as focal cortical dyspla-
sia [67]. Some ECoG findings that were thought to be specific to fo-
cal cortical dysplasia [67] were obtained in other pathologies [68]
and, therefore, seem not to be as specific as previously thought.
4. How good is therapy?
Where the usefulness of EEG as an aid to diagnosis of seizure
disorders is established, its role as a guide to monitoring treatment
is much less certain [69]. The EEG may reflect unspecific central
nervous side effects of the standard AEDs such as carbamazepine,
valproate, phenytoin, benzodiazepines, and phenobarbital [70].
All standard AEDs may result in slowing of the dominant rhythm
and increased slow activity. Carbamazepine, in particular, is often
associated with apparent deterioration of background activity,
even in the face of clinical improvement. Less information is avail-
able on the newer AEDs [71]. Benzodiazepines and barbiturates re-
sult in increased fast activity [69]. Intravenous benzodiazepines
and phenytoin result in both acute seizure control and suppression
of IEDs.
There is a very close correlation between control of clinically
detected seizures and EEG events for patients with absence epi-
lepsy and 3-Hz SWCs [69]. In some, but not all, patients with other
seizure disorders, there is a positive correlation between number
of seizures and amount of IEDs. For seizures other than absences,
and AEDs given in the medium and long term, there is generally
not a clear relationship between control of seizures and IEDs
[69]. In studies of children whose epilepsy is in remission, persis-
tent IEDs have been associated with a higher risk of seizure relapse
should AEDs be discontinued, but in adults the relevance of persis-
tent IEDs appears to be much less certain. The risk for seizure re-
lapse is increased if the EEG showed the slowing prior to
medication withdrawal (relative risk = 2.4) [72].
32 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33
Postoperative routine EEG is a good prognostic instrument for
the prediction of long-term seizure outcome following anterior
temporal resection. It predicts the recurrence of single seizures in
patients with rare seizures [73].
Repeated EEG in patients with an established diagnosis of epi-
lepsy was useful in the aforementioned settings.
References
[1] Engel Jr J. A practical guide for routine EEG studies in epilepsy. J Clin
Neurophysiol 1984;1:109–42.
[2] Lüders HO, Noachtar S. Atlas and classification of electroence-
phalography. Philadelphia: Saunders; 2000.
[3] Noachtar S, Binnie C, Ebersole J, et al. For the International Federation of
Clinical Neurophysiology. A glossary of terms most commonly used by clinical
electroencephalographers and proposal for the report form for the EEG
findings. Electroencephalogr Clin Neurophysiol Suppl 1999;52:21–41.
[4] Daly DD. Epilepsy and syncope. In: Daly DD, Pedley TA, editors. Current
practice of clinical electroencephalography. New York: Raven Press; 1990. p.
269–334.
[5] Jabbari B, Russo MB, Russo ML. Electroencephalogram of asymptomatic adult
subjects. Clin Neurophysiol 2000;111:102–5.
[6] Gregory RP, Oates T, Merry RT. Electroencephalogram epileptiform
abnormalities in candidates for aircrew training. Electroencephalogr Clin
Neurophysiol 1993;86:75–7.
[7] Eeg-Olofsson O, Petersen I, Sellden U. The development of the
electroencephalogram in normal children from the age of 1 through 15
years: paroxysmal activity. Neuropädiatrie 1971;4:375–404.
[8] Kellaway P. The electroencephalographic features of benign centrotemporal
(rolandic) epilepsy of childhood. Epilepsia 2000;41:1053–6.
[9] Lüders HO, Lesser RP, Dinner DS, et al. Benign focal epilepsy of childhood. In:
Lüders H, Lesser RP, editors. Epilepsy: electroclinical
syndromes. London: Springer; 1987. p. 303–46.
[10] Zivin L, Marsan CA. Incidence and prognostic significance of ‘‘epileptiform”
activity in the EEG of non-epileptic subjects. Brain 1968;91:751–78.
[11] Ajmone-Marsan CA, Zivin LS. Factors related to the occurrence of typical
paroxysmal abnormalities in the EEG records of epileptic patients. Epilepsia
1970;11:361–81.
[12] Van Donselaar CA, Schimsheimer RJ, Geerts AT, et al. Value of the
electroencephalogram in adult patients with untreated idiopathic first
seizures. Arch Neurol 1992;49:231–7.
[13] Goodin DS, Aminoff MJ. Does the interictal EEG have a role in the diagnosis of
epilepsy? Lancet 1984;1:837–9.
[14] Salinsky M, Kanter R, Dasheiff RM. Effectiveness of multiple EEGs in
supporting the diagnosis of epilepsy: an operational curve. Epilepsia
1987;28:331–4.
[15] Fountain NB, Kim JS, Lee SI. Sleep deprivation activates epileptiform discharges
independent of the activating effects of sleep. J Clin Neurophysiol
1998;15:69–75.
[16] Fountain NB, Freeman JM. EEG is an essential clinical tool: pro and con.
Epilepsia 2006;47(Suppl. 1):23–5.
[17] Hauser WA, Rich SS, Annegers JF, et al. Seizure recurrence after a 1st
unprovoked seizure: an extended follow-up. Neurology 1990;40:1163–70.
[18] Williams GW, Luders HO, Brickner A, et al. Interobserver variability in EEG
interpretation. Neurology 1985;35:1714–9.
[19] Matsuo F, Knott JR. Focal positive spikes in electroencephalography.
Electroencephalogr Clin Neurophysiol 1977;42:15–25.
[20] Gloor P. Consciousness as a neurological concept. Epilepsia 1986;27:S14–26.
[21] Noachtar S. Der Klicker-Test: eine einfache Methode zur Überprüfung und
Dokumentation der Bewusstseinslage im EEG. EEG Labor 1993;15:41–6.
[22] Lüders HO, Awad IA. Conceptual considerations. In: Lüders HO, editor. Epilepsy
surgery. New York: Raven Press; 1992. p. 51–62.
[23] Gibbs FA, Davis H, Lennox WG. The EEG in epilepsy and impaired states of
consciousness. Arch Neurol Psychiatry 1935;34:1133–48.
[24] Browne TR, Penry JK, Porter RJ, et al. Responsiveness before, during and after
spike-wave paroxysms. Neurology 1974;24:659–65.
[25] Wolf P, Goosses R. Relation of photosensitivity to epileptic syndromes. J Neurol
Neurosurg Psychiatry 1986;49:1386–91.
[26] Leutmezer F, Lurger S, Baumgartner C. Focal features in patients with
idiopathic generalized epilepsy. Epilepsy Res 2002;50:293–300.
[27] Lombroso CT. Consistent EEG focalities detected in subjects with primary
generalized epilepsies monitored for two decades. Epilepsia 1997;38:
797–812.
[28] Gastaut J, Roger J, Soulayrol R, et al. Childhood epileptic encephalopathy with
diffuse slow spike–waves (otherwise known as ‘petit mal variant’) or Lennox
syndrome. Epilepsia 1966;7:139–79.
[29] Lennox WG, Davis JP. Clinical correlates of the fast and slow spike wave
electroencephalogram. Pediatrics 1950;5:626–44.
[30] Noachtar S, Winkler PA, Lüders HO. Surgical therapy of epilepsy. In: Brandt T
et al., editors. Neurological disorders: course and treatment. San
Diego: Academic Press; 2003. p. 235–44.
[31] Gibbs EL, Gibbs FA, Fuster B. Psychomotor epilepsy. Arch Neurol Psychiatry
1948;60:331–9.
[32] Semah F, Picot MC, Adam C, et al. Is the underlying cause of epilepsy a major
prognostic factor for recurrence? Neurology 1998;51:1256–62.
[33] Kanner AM, Parra J, Gil-Nagel A, et al. The localizing yield of sphenoidal and
anterior temporal electrodes in ictal recordings: a comparison study. Epilepsia
2002;43:1189–96.
[34] Morris 3rd HH, Kanner A, Lüders H, et al. Can sharp waves localized at the
sphenoidal electrode accurately identify a mesio-temporal epileptogenic
focus? Epilepsia 1989;30:532–9.
[35] Schulz R, Lüders HO, Hoppe M, et al. Interictal EEG and ictal scalp EEG
propagation are highly predictive of surgical outcome in mesial temporal lobe
epilepsy. Epilepsia 2000;41:564–70.
[36] Pfänder M, Arnold S, Henkel A, et al. Clinical features and EEG findings
differentiating mesial from neocortical temporal lobe epilepsy. Epileptic
Disord 2002;4:189–95.
[37] Krendl R, Lurger S, Baumgartner C. Absolute spike frequency predicts surgical
outcome in TLE with unilateral hippocampal atrophy. Neurology 2008;71:
413–8.
[38] Rémi J, Heinlin J, De Marinis A, et al. Correlation of interictal and ictal EEG
findings to MRI-documented lesions in patients with drug-resistant focal
epilepsies. In: Abstracts, 2008 annual meeting of the Epilepsy Society,
Saturday, December 6, Poster Session 1. Epilepsia 2008;49(Suppl. 7):1–166.
[39] Noachtar S, Bilgin O, Remi J, et al. Interictal regional polyspikes in noninvasive
EEG suggest cortical dysplasia as etiology of focal epilepsies. Epilepsia
2008;49:1011–7.
[40] Kilpatrick C, Cook M, Kaye A, et al. Non-invasive investigations successfully
select patients for temporal lobe surgery. J Neurol Neurosurg Psychiatry
1997;63:327–33.
[41] Sperling MR, O’Connor MJ, Saykin AJ, et al. A noninvasive protocol for anterior
temporal lobectomy. Neurology 1992;42:416–22.
[42] Winkler PA, Herzog C, Henkel A, et al. Noninvasive protocol for surgical
treatment of focal epilepsies. Nervenarzt 1999;70:1088–93.
[43] Foldvary N, Klem G, Hammel J, et al. The localizing value of ictal EEG in focal
epilepsy. Neurology 2001;57:2022–8.
[44] Reiher J, Beaudry M, Leduc CP. Temporal intermittent rhythmic delta activity
(TIRDA) in the diagnosis of complex partial epilepsy: sensitivity, specificity
and predictive value. Can J Neurol Sci 1989;16:398–401.
[45] Beleza P, Bilgin O, Noachtar S. Interictal rhythmical midline theta differentiates
frontal from temporal lobe epilepsies. Epilepsia 2009;50:550–5.
[46] Ciganek L. Theta-discharges in the middle-line: EEG symptom of temporal lobe
epilepsy. Electreoencephaolgr Clin Neurophysiol 1961;13:669–73.
[47] Winkler PA, Herzog C, Henkel A, et al. Nicht-invasives Protokoll für die
epilepsiechirurgische Behandlung fokaler Epilepsien. Nervenarzt
1999;70:1088–93.
[48] Noachtar S. Subdural electrodes in focal cortical dysplasia. Epileptic Disord
2003;5(Suppl. 2):S91–4.
[49] Van Buren JM. Complications of surgical precedures in the diagnosis and
treatment of epilepsy. In: Engel JJ, editor. Surgical treatment of the
epilepsies. New York: Raven Press; 1987. p. 465–75.
[50] So N, Gloor P, Quesney LF, et al. Depth electrode investigations in patients with
bitemporal epileptiform abnormalities. Ann Neurol 1989;25:423–31.
[51] So N, Olivier A, Andermann F, et al. Results of surgical treatment in patients
with bitemporal epileptiform abnormalities. Ann Neurol 1989;25:432–9.
[52] Lee SK, Yun CH, Oh JB, et al. Intracranial ictal onset zone in nonlesional lateral
temporal lobe epilepsy on scalp ictal EEG. Neurology 2003;61:757–64.
[53] Duncan JS. Imaging and epilepsy. Brain 1997;120:339–77.
[54] Arnold S, Berthele A, Drzezga A, et al. Reduction of benzodiazepine receptor
binding is related to the seizure onset zone in extratemporal focal cortical
dysplasia. Epilepsia 2000;41:818–24.
[55] Weil S, Noachtar S, Arnold S, et al. Ictal ECD-SPECT differentiates between
temporal and extratemporal epilepsy: confirmation by excellent postoperative
seizure control. Nucl Med Commun 2001;22:233–7.
[56] Noachtar S, Arnold S, Yousry TA, et al. Ictal technetium-99m ethyl cysteinate
dimer single-photon emission tomographic findings and propagation of
epileptic seizure activity in patients with extratemporal epilepsies. Eur J
Nucl Med 1998;25:166–72.
[57] Morris HH. Frontal lobe epilepsies. In: Lüders HO, editor. Epilepsy
surgery. New York: Raven Press; 1991. p. 157–65.
[58] Gates JR, Leppik IE, Yap J, Gumnit RJ. Corpus callosotomy: clinical and
electroencephalographic effects. Epilepsia 1984;25:308–16.
[59] Tellez-Zenteno JF, Dhar R, Wiebe S. Long-term seizure outcomes following
epilepsy surgery: a systematic review and meta-analysis. Brain 2005;128:
1188–98.
[60] Holthausen H, Noachtar S, Pannek H, et al. Foramen ovale and epidural peg
electrodes. Acta Neurol Scand Suppl 1994;152:39–43.
[61] Noachtar S. Epidural electrodes. In: Lüders HO, Comair Y, editors. Epilepsy
surgery. New York: Lippincott Raven; 2001.
[62] Noachtar S, Holthausen H, Sakamoto A, et al. Semi-invasive Elektroden in der
epilepsiechirurgischen Diagnostik. In: Stefan H, editor. Epilepsie
92. Berlin: Deutsche Sektion der Internationalen Liga gegen Epilepsie; 1993.
p. 148–52.
[63] Awad IA, Assirati Jr JA, Burgess R, et al. A new class of electrodes of
‘intermediate invasiveness’: preliminary experience with epidural pegs and
foramen ovale electrodes in the mapping of seizure foci. Neurol Res
1991;13:177–83.
[64] Wieser HG, Elger CE, Stodieck SR. The ‘foramen ovale electrode’: a new
recording method for the preoperative evaluation of patients suffering from
 S. Noachtar, J. Rémi / Epilepsy & Behavior 15 (2009) 22–33 33

mesio-basal temporal lobe epilepsy. Electroencephalogr Clin Neurophysiol [69] Duncan JS. Antiepileptic drugs and the electroencephalogram. Epilepsia
1985;61:314–22. 1987;28:259–66.
[65] Schuler P, Neubauer U, Schulemann H, et al. Brain-stem lesions in the course of [70] Bauer G. EEG, drug effects, and central nervous system poisoning. In:
a presurgical re-evaluation by foramen-ovale electrodes in temporal lobe Niedermeyer E, Lopes da Silva FH, editors. Electroencephalography: basic
epilepsy. Electroencephalogr Clin Neurophysiol 1993;86:301–2. principles, clinical applications, and related fields. Baltimore: William &
[66] Fernandez G, Hufnagel A, Van Roost D, et al. Safety of intrahippocampal depth Wilkins; 1993. p. 631–42.
electrodes for presurgical evaluation of patients with intractable epilepsy. [71] Van Gott AC, Brenner RP. Drug effects and toxic encephalopathies. In: Ebersole
Epilepsia 1997;38:922–9. J, Pedley TA, editors. Current practice of clinical
[67] Palmini A, Gambardella A, Andermann F, et al. Intrinsic epileptogenicity of electroenecephalography. Philadelphia: Lippincott; 2003. p. 463–82.
human dysplastic cortex as suggested by corticography and surgical results. [72] Shinnar S, Berg AT, Moshe SL, et al. Discontinuing antiepileptic drugs in
Ann Neurol 1995;37:476–87. children with epilepsy: a prospective study. Ann Neurol 1994;35:534–45.
[68] Rosenow F, Luders HO, Dinner DS, et al. Histopathological correlates of [73] Hildebrandt M, Schulz R, Hoppe M, et al. Postoperative routine EEG correlates
epileptogenicity as expressed by electrocorticographic spiking and seizure with long-term seizure outcome after epilepsy surgery. Seizure
frequency. Epilepsia 1998;39:850–6. 2005;14:446–51.