Electrocorticography (ECOG) is a neurophysiological technique used to
record cortical potentials from the exposed brain in the operating room.
It has been in common use for over four decades in the surgical treatment
of people with medically refractory epilepsy [1, 2]. This paper revisits
the role of ECOG. The main reason for recording an ECOG has been to confirm
and outline the actual site and extent of the epileptogenic process prior
to resection. After the resection has been completed, ECOG has been used
to determine if all the potentially epileptogenic tissue has been removed.
Both of these roles have been questioned.
The ECOG records the same type of cerebral potentials as the scalp electroencephalogram,
except that the dispersion and attenuation of the potential by the scalp
and skull is not present. In theory, this should allow for better localization
of the origin of the epileptogenic tissue causing the patient's habitual
seizures. Because of the short recording time available for ECOG and the
limitations of electrode placement posed by the surgical exposure, the
possibility of recording ictal events is rare. The ECOG relies on the
presence of interictal epileptiform discharges for the identification
of the irritative/epileptic zone. Careful preoperative planning is required.
The clinical history, electroencephalographic studies (both ictal and
interictal), and imaging studies need to be carefully reviewed to define
the area of resection and exposure needed for adequate ECOG recording
prior to surgery.
The actual technique of recording an ECOG has changed little since its
introduction. Throughout the time of the recording, continuous close communication
between the clinical neurophysiologist and the neurosurgeon is required.
The clinical neurophysiologist must be able to read and interpret the
recording as it is being done. Any modifications of electrode position
and montages must be done at that time. This requires that the recording
apparatus be either in the operating room or in the operating room gallery
with a two way communication system in place.
Standard sixteen channel EEG equipment can give very adequate ECOG recordings.
As ECOG records directly from the brain, modifications in recording sensitivities,
filters, and time constants will be required. Electrodes need to be able
to sit on the leptomeninges and move with the pulsations of the brain.
Either flexible ball electrodes mounted on a fixed horseshoe frame or
a series of electrodes implanted in a soft flexible silastic grid can
be used. The advantage of the ball electrodes over strip electrodes is
that they can be sterilized and reused, cutting down costs. Due to the
lack of adequate sterilization techniques, silastic implanted electrode
should be used only once.
For accurate localization of the epileptogenic area, the electrodes
need to be placed an equal distance apart. Montages should consist of
a minimum of four electrodes in a straight chain. The use of montages
using angulated electrode placement should be avoided as this can lead
to false localization. If simultaneous recording from multiple chains
of electrodes is planned, the electrodes must be of equal distance in
both vertical and horizontal planes. This will allow for bipolar recordings
to be done, as well as, referential recordings. When single chains of
electrodes are used only, both referential and bipolar recordings should
be done simultaneously. Various referential electrode placements can be
used (such as mastoid, cervical region, and bone flap).
To record from the deeper structures (such as the mesial temporal regions)
either four contact
depth electrode can be inserted through the brain to reach the deeper
structure or a flexible silastic four-electrode grid can be inserted under
the temporal/frontal region. The placement of a chain of electrodes over
the cortical surface at the same time as the recording allows for a better
understanding of the propagation of the abnormal cerebral potentials.
Total recording time is about thirty minutes, although longer times
are occasionally necessary. Methohexital and alfentanil have been reported
to "activate the epileptic zone" [3, 4]. These techniques should be done
only after adequate standard recordings of the exposed brain have been
done. Careful interpretation of the areas of "activation" must be made
as the potential to "activate" regions outside the true irritative/epileptic
zones has been reported.
In adults, the ECOG recording is usually done with the patient in the
awake state; however, in children this is not often possible. The anaesthetic
agents used for the operative procedure can have significant effects on
the ECOG. Sufentanil, fentanyl, alfentanil, propofil methohexital have
been reported to produce epileptiform changes on ECOG, whereas halothane,
barbituates and benzodiazepines may suppress the epileptic activity [3-7].
Both nitrous oxide and isoflurane have been reported to affect the ECOG,
but more recent reports have suggested that this is not the case when
low concentrations are used [8, 9]. For these reasons, it is recommended
that in patients be maintained on nitrous oxide, isoflurane and non-depolarizing
muscle relaxants be used during ECOG recording.
During the ECOG, electrical stimulation of the cortex has been used
to localize the area of epileptic seizure onset, as well as to map the
area of cortex responsible for motor, sensory and language functions.
While the ECOG is being recorded, an electrical stimulus is delivered
between two ball electrodes or between adjacent silastic enclosed electrodes.
This stimulus consists of a 0.5 to 2 nsec diphasic pulse applied at 50-60
Hz over 1 to 5 seconds with an intensity of 0.5 to 2.0 mA and a voltage
of 1 to 15 V. The site of the patient's habitual seizure aura is noted,
as well as the regions of maximal after-discharges. These discharges are
in the form of high frequency stereotypic paroxysms such as spikes, rhythmical
sharp waves or spike wave sequences. The discharge usually ends abruptly.
It may be localized or spread to multiple electrode sites.
In order to examine the validity of ECOG, a Medline literature search
from 1985 to the present was performed using the key words electrocorticography,
ECOG, intra-operative recording, and epilepsy. As complete coverage of
the medical literature for a given topic using Medline is often incomplete,
a search of the references quoted in each of the articles was made. The
data were reviewed by type of resection (i.e. temporal, extra-temporal
or lesional) and the degree of evidence in support or against the use
of ECOG for each of the previously stated reasons.
Nonlesional temporal lobe resections
According to Rasmussen , the removal of the "epileptogenic zone"
(the anatomical site of seizure onset), as well as the surrounding tissue
which might potentially be recruited into the critical mass of tissue,
is required for a successful surgical outcome. ECOG has been reported
to provide useful information for the localization of this area at the
time of surgery [11-15].
Thirty-two papers on this topic were located by our search technique
The review of these papers revealed little evidence that actually established
the true value of the procedure in this role. All that have been published
have been in the form of retrospective case studies or case series without
a proper control group. Some investigators have questioned the role of
this procedure .
Epileptiform discharges were reported to have been recorded from the
temporal lobe in all of the papers located by our search mechanism. These
waveforms were similar to those recorded from the scalp electroencephalogram,
consisting of isolated spikes, brief bursts of spikes or runs of sharp
waves . They were often multifocal with a wide distribution over the exposed
temporal cortex. More commonly, the epileptiform discharges were reported
to have been recorded from the hippocampal structures and inferiomesial
surfaces of the temporal tip. To a much lesser extent , they were recorded
from the lateral temporal cortices and more so from the posterior temporal
cortices [15, 17-21]. Propagation of the epileptiform discharge was commonly
seen from the hippocampus, to and from the subtemporal cortex [19, 21].
Parent-Raymond et al.  have suggested that in addition to
these epileptiform discharges, delta wave activity might also give useful
information as to the location of the epileptiform tissue in patients
in whom spikes were not found on ECOG. With the help of spectral analysis
of the ECOG data, the maximum area of delta frequency activity was determined
in forty patients. In twenty-two patients, this area coincided with the
region of maximum spike activity. This association was statistically more
frequent than would have been predicted. The authors do suggest caution
when using this method in individual cases.
In theory, ECOG recordings could be expected to provide information
missed on a scalp electroencephalogram especially activity from the mesial
surface of the temporal lobe. Few studies have actually addressed this
question. Devinsky et al.  in a report of a retrospective review
of thirty-three patients with medically refractory epilepsy who underwent
temporal lobectomy, found eight patients in whom the ECOG recorded epileptiform
discharges not previously found by scalp electroencephalograms even with
the use of sphenoidal electrodes. On the other hand, Engel et al.
 noted that no additional information was obtained from the ECOG compared
to the preoperative scalp electroencephalogram.
ECOG seldom records ictal events, but rather interictal epileptiform
activity. The relationship between the epileptic zone (area of origin
of the epileptic seizure) and the irritative zone (area of maximum interictal
epileptiform discharge) is not completely understood. In particular, the
degree to which the two zones overlap, especially on the ECOG recording
is unclear. It has been hypothesized that the more frequent these discharges
are within an area, the more likely it is that this area lies within the
epileptic zone . Alarcon et al.  found that the removal
of this area resulted in a significantly better chance of a good surgical
outcome, even if areas of less frequent discharging were left untouched.
If the area of maximal discharging was not completely resected, the surgical
outcome was more likely to be poor. An earlier study by Tran et al.
 had found no association between surgical outcome and frequency of
epileptiform discharges. This study used the visual analysis of the ECOG
record by the electroencephalographer, whereas Alarcon had used a computerized
spike detection program.
Electrical stimulation of the cortex at the time of ECOG has been used
in further localization of the epileptic zone [1, 13]. The site of reproduction
of the habitual aura of the patient by electrical cortical simulation
has been reported by some investigators to have strong correlation with
the epileptic zone [26-28]. This was particularly the case when the area
from which the responses were elicited also coincided with the region
exhibiting the greatest epileptic discharge. Gloor  commented that
this was less likely to be the case if an afterdischarge occurred, especially
it if spreads to distant areas. In this situation the clinical symptomatology
may not be related to this area of stimuli, but reflect a distant area
to which the discharge had spread.
After-discharges arising from electrical cortical stimulation are of
questionable value in the localization of the epilep-
tic zone. It is not uncommon for after
discharge to issue from regions of the brain from which no epileptiform
abnormalities had been recorded on ECOG [1, 13, 29]. The thresholds for
has been reported to show considerable variability .
New epileptiform discharges can appear after the excision; when along
the margin of the resection, they may be representing a cortical disturbance
caused by the manipulation of the brain at the time of surgery and may
not have prognostic significance. Discharges some distance from the resection
margin may be more predictive of a poor surgical outcome. These differences
may explain the difference in the reported use of postresection ECOG in
predicting surgical outcome. This cannot be confirmed from the reported
studies as the area of residual discharge in relation to the resection
margin was often unclear.
The role of the ECOG in determining the surgical outcome is unclear.
McBride et al.  and Engel et al.  did not find an
association between the site of dominant intraoperative ECOG epileptiform
discharge and the surgical outcome. Correlation between the location of
discharge and underlying pathology was also not found .
Bengzon et al.  noted a significant difference in the surgical
outcome between patients with residual spikes compared to spike-free postresection
recordings. Thirty-six percent of patients who were seizure-free postsurgery
had residual spike; whereas 75% of patients who were not seizure-free
had residual spikes on post-resection ECOG. These findings were similar
to those reported by Jasper , Foil et al. , Wyllie et
al. , Drake et al. , Tanaka et al.  and
McBride et al. .
Other investigators have not been able to confirm the relationship between
the degree of epileptiform discharges seen on the post-resection ECOG
and the outcome [16, 36-39]. This was especially the case in patients
in whom selective amygdalo-hippocampectomy was done. In this situation,
new epileptiform discharges were often recorded from the temporal cortex
after completion of the procedure. These discharges were found to have
no predictive value [40, 41].
A possible explanation for the lack of consensus as to the role of the
ECOG in the prediction of surgical outcome may lie in the surgical procedure
itself. Most centres that have reported the use of ECOG-performed standardized
temporal lobe resections. Only a few centres actually tailored the resection
according to ECOG findings .
Evidence in support of the use of ECOG in extra-temporal resections
is scant (six papers). Most of the literature pertaining to this topic
is related to lesional cases, and will be covered in that section. Quesney
et al. [43-45] noted that in patients whose seizures originated
in the frontal lobes, there was no clear relationship between the amount
of epileptogenic tissue removed and a successful surgical outcome. The
reason for this may lie in the anatomical and functional peculiarities
of this region, which permits the epileptiform discharge to vary in location
from a specific region, to multilobular or even bifrontal. Similar findings
have been noted in ECOG recordings of patients with seizures arising from
the centroparietal region and occipital areas [43, 44].
For patients with non-lesional, medically refractory extra-temporal
epilepsy, the use of subdural grid electrodes may prove to be the answer
to this question. As this procedure allows ictal events to be recorded,
there is a greater possibility that the epileptic zone will be identified.
Another approach may involve the use of other imaging techniques, for
example positron emission tomography .
Lesional, temporal and
The use of ECOG in patients in whom structural lesions have been identified
on imaging studies remains controversial. Traditional wisdom based on
the Montreal Neurological Institute experience holds that optimum seizure
control is achieved when the lesion is removed with the surrounding epileptogenic
cortex as determined by ECOG [45-48]. This point of view has been supported
by Pilcher et al.  who found that eleven out of twelve patients
who underwent surgery for ganglioglioma, were seizure-free at 3.1 years
post-surgery compared with a literature control of twenty-one out of thirty-nine
(54%) patients with ganglioglioma in whom only the lesion was resected.
They noted that the epileptogenic zone was topographically distinct from
the region of the tumour-involved brain. It usually encompassed a large
surface area. Non-epileptiform high-amplitude slow waves were recorded
on ECOG predominantly in tumour-involved cortex, while epileptiform spike
discharges were recorded over normal-appearing cortex. They felt that
these discharges represented the epileptogenic zone. The removal of this
area increased the chance of a better outcome.
Berger et al.  reported a series of children and adults with
intractable epilepsy, associated with low grade tumour. Forty-one out
of forty-five of these patients (91%) became seizure-free. They did not
advocate the use of ECOG in patients with lesion and occasional or new
onset of seizures, but felt it should be used in individuals with medically
refractory epilepsy associated with low grade tumour. This was particularly
the case in children, where there was a significantly better chance of
seizure control. Cohen et al.  reported similar findings in
98 patients with cavernous hemangioma. Gonzalez et al.  also
reported greater control of seizures when the tumour was resected with
ECOG-guided removal of the seizure focus. Drake et al.  reported
the results of a series of children with structural lesions of the temporal
lobe using ECOG guidance. They again emphasized the concept that seizure
activity often originated from the brain tissue adjacent to the tumour.
Others have not felt that ECOG guidance is necessary in this group of
patients. Resections have been restricted to the tumour margins as delineated
on imaging studies and at the time of surgery [53-55]. The literature
states that if the lesion has not been completely removed, seizures will
continue despite the removal of the "epileptogenic zone" [56, 57].
These studies have been based on retrospective review of case series.
Comparison between ECOG-guided resections and non-ECOG-guided resection
for similar groups of lesions have not been prospectively done. Tran et
al.  attempted a controlled retrospective series review in which
patients with structural lesions present underwent resection of the lesion
to normal tissue margins. ECOG was recorded pre- and post-resection, but
not used to determine the amount of surgical resection. Patient outcome
was based upon seizure-free state. ECOGs were analysed for spike distribution
and spike discharge rate. Spikes were found to be over the tumour bed
as well as in the surrounding tissues. Spike distribution in the pre-resection
ECOG did not correlate with outcome. On post-resection ECOG, spikes were
noted along the edge of the resection as well as extra-marginally in equal
amounts between patients who became seizure-free post-resection and those
who did not. These findings support the use of lesionectomy as the first
step in seizure control in lesional cases. Like temporal cortical resections,
the presence of post-resection spikes does not appear to correlate with
outcome. A case may be made for the use of ECOG in patients with dual
pathology to assess the degree of epileptogenesis in the distant site
. Clarke et al.  have suggested that even in this case,
the lesion must be removed if good seizure control is to be obtained.
Cortical dysplastic lesions are often associated with severe, partial
epilepsies of childhood and can prove refractory to medical management
leading to the need for surgical resection. Palmini et al. 
have reported long runs of epileptiform discharges on ECOG consisting
of repetitive electrographic seizures, repetitive bursting discharges
or continuous rhythmic spiking. They felt that these discharges were often
co-localized with the magnetic resonance imaging-defined lesion. The completeness
of resection of the epileptiform activity on ECOC correlated with the
surgical outcome. Wennberg et al.  noted similar findings in
lesion-related frontal lobe resection. This would suggest a role for ECOG
in these cases to determine the location and extent of resection at the
time of surgery. Holmes et al.  caution that neocortical lesions
on magnetic resonance imaging (MRI) do not necessarily indicate the site
of ictal onset of partial epilepsy. They reported on twenty adult patients
with medically refractory epilepsy where the MRI showed a focal lesion.
Electroencephalogram suggested that ictal onsets were in different areas.
Surgery was based on electrographic data. Fifty percent of patients were
seizure free at follow up. Thirty-five percent had a greater than 75%
reduction in their seizures.
For several decades, ECOG recordings have been routinely used in the
surgical management of patients with medically refractory epilepsy. Using
this technique, confirmation of epileptiform activity on pre-operative
scalp electroencephalogram has usually been demonstrated at the time of
surgery. It also allows for direct exploration and recording of the mesial
surfaces of the cerebral cortex. With this information, tailored resections
of epileptogenic tissue can be performed. Whether tailored resections
have better outcome than standard resections of the temporal lobe, remains
unclear. In order to answer this question, carefully designed, controlled
prospective study will need to be done, controlling for variables such
as the type of resection, location and degree of epileptiform disturbance
on ECOG, and outcome measures such as seizure control and psycho-social
status. It is quite possible that the results of ECOG-guided and standard
temporal resection will be very similar, when the size and extent of the
surgical resection will turn out to be the same in both groups.
The role of ECOG in extratemporal and lesional resection is different
in the case of temporal resection. The ECOG appears to be beneficial in
lesional resections in patients with refractory epilepsy. Its use in new
onset cases is less clear. Also its use in patients with dual pathology
remains unclear. Controlled prospective studies again would be helpful.
Extratemporal resections need to be well planned prior to surgery. Imaging
studies such as MRI and PET have proven helpful in localizing the site
of epileptogenicity. The degree of electrographic spread of epileptiform
discharges suggests that chronic intracranial recordings may be needed
to provide more accurate and complete information about the seizure origin
than intraoperative ECOG.