Reports and experience worldwide have shown that resection of the epileptogenic area of brain in uncontrolled epilepsy patients with stereotyped partial seizures can cure the epilepsy, or at least prevent recurrent seizures.^30,57,81 The corollary of this statement is that if the surgery fails, the area removed was incorrect or incompletely resected or was only one of several cerebral regions responsible for seizures. Accordingly, accurate definition of the location and limits of this epileptogenic area has become a major preoccupation among epileptologists.

The first need is to define epileptogenic area, a term synonymous with epileptogenic region or epileptogenic zone. This is defined as the area of brain necessary to generate seizure activity,³¹ resection of which (or inactivation or complete disconnection of which) will result in cessation of seizures. We seek to differentiate this epileptogenic zone from the irritative zone (of interictal spiking), the ictal onset zone (where the electroencephalogram [EEG] shows spontaneous seizures to arise), the symptomatogenic zone (activation of which produces initial ictal symptoms or signs), the zone of functional deficit (with evidence of reduced or abnormal metabolism, blood flow, and cognitive ability), and the epileptogenic lesion (demonstrated by structural imaging).⁷⁴ Although those areas usually overlap, they do not precisely coincide with the epileptogenic zone and, at times, appear to diverge. We might eventually develop techniques that allow visualization of the epileptogenic zone in other ways, but currently the demonstration of electrographic seizure onset offers the most logical means of verifying its approximate location,³⁶ except in childhood catastrophic epilepsy, in which the ictal onset zone is typically reported to be generalized even if the causative, epileptogenic zone is localized.⁷² (An exception to this is the occasional case of childhood catastrophic epilepsy in which intracranial recording has demonstrated intrinsic epileptogenicity of a lesion, such as hypothalamic hamartoma.)⁴⁸ Because the latter cases are not likely to be studied with intracranial electrodes, this chapter concentrates almost exclusively on the evaluation of partial seizure disorders.

The electrographic delineation of the epileptogenic zone can be supported by delineation of the irritative zone, the symptomatogenic zone, the area of focal functional deficit, and the epileptogenic lesion such that the complementary nature of the respective studies allows one to infer the limits of the epileptogenic zone from less precise electrographic information, a scenario most often found when an epileptogenic lesion is roughly colocalized with a surface EEG recording abnormality.^32,33,83 In other situations, however, particularly in the absence of compelling data to characterize the epileptogenic lesion, irritative zone, and zone of functional deficit, the ability of the surface EEG to determine the ictal onset zone is suboptimal. For the purpose of designing an effective resective procedure for uncontrolled epilepsy, intracranial electrodes are then used to localize ictal onset as well as interictal spike distribution; this information is then used with the other data to infer the location of the epileptogenic area.

Intracranial EEG recording fills a specific need only in the context of a specific question. Extensive knowledge of seizure patterns, surface EEG expression of the ictal onset and irritative zones, and functional and structural abnormalities accompanying the epilepsy is necessary to properly design an intracranial EEG study. The nature of the procedure requires that its risk be sufficiently balanced by the possibility of useful gain in the localization, definition, and understanding of the uncontrolled epilepsy. In addition, the appropriate intracranial techniques must be used to answer the questions posed.

Intracranial EEG is indicated in the setting of uncontrolled epilepsy considered for resective surgical treatment when (a) there is likely to be a single region of seizure onset; (b) localization of this region is insufficiently precise or impossible based on the combination of ictal clinical manifestations, noninvasive EEG, and tests of focal functional deficit and structural abnormalities; and (c) the spectrum of findings in the individual allows a hypothesis about the suspected region or regions of seizure onset to be further investigated.

These considerations are important because intracranial recording has significant limitations, costs, and risks.⁹¹ Implanted electrodes necessarily record from a limited volume of cortex. They must therefore be placed appropriately to address the questions at hand. Furthermore, because chronic implantation carries the known risks of hemorrhage and infection and the unknown risks of cognitive effects and epileptogenic potential, it is desirable to restrict the number of implanted electrodes to as few as possible. This must be done without compromising the ability to detect the zone of seizure onset with sufficient precision to use in the determination of the epileptogenic area for subsequent surgery.

Technological advances have altered the use of intracranial recording over the past decade in several ways. High-quality neuroimaging, with qualitative and quantitative magnetic resonance imaging (MRI) at the forefront, has improved our detection of macrostructural abnormalities in epilepsy, including certain gliomas, developmental abnormalities (cortical dysplasia, hamartomas), and restricted atrophic lesions such as hippocampal atrophy. As accumulated experience has supported significant correlations between excision of these lesions and seizure relief, finding evidence for a structural lesion has assumed increasing importance in the localization of the epileptogenic zone for epilepsy surgery.^16,20,83,89 Now, the presence of certain “lesions” more often allows the less precise scalp EEG localization of the ictal onset zone to suffice because the additional information required to infer the localization of the epileptogenic area is provided by the MRI. Although scalp EEG localization to the region of the structural abnormality is still considered imperative to prove that the lesion is epileptogenic,
localization by interictal surface EEG without discordant information from ictal surface recording has become acceptable for this purpose at most centers. The opinion that only imaging might be sufficient to define a cortical region of epileptogenesis has been voiced, but some patients are known to have coincidental structural lesions irrelevant to the electrically abnormal area.^34,84 In Yale’s recent experience, patients with discordant MRI and overall EEG localization represent approximately 10% of those with hippocampal atrophy, tumors, or developmental lesions.³⁴ About half of these patients have had surgery based on ictal onset localization, with cure of seizures. This means that, although the MRI is highly sensitive in localizing epileptogenic structural lesions in uncontrolled epilepsy patients and these lesions are often where uncontrolled seizures begin, the structural abnormality does not necessarily indicate the epileptogenic zone.

The increasing sophistication of imaging has also been responsible for generating a new population of patients for intracranial study who may not have been considered candidates before. These are patients with bilateral hippocampal atrophy or with so-called “dual pathology” such as a tumor or cortical developmental abnormality with hippocampal atrophy. In these situations, experts agree that intracranial recording is essential to determine which potentially epileptogenic lesion is indeed responsible for the uncontrolled seizure disorder.³⁴ In one recent study, patients with bilateral hippocampal atrophy were found to be good candidates for cure of seizures after temporal lobe resection, with the side of resection based on intracranial ictal onset EEG localization and not on the severity of the atrophy.⁵¹ An MRI demonstration of dual pathology is also not always synonymous with dual epileptogenesis, although it might be.

Intracranial EEG has been in use since the 1950s. It was initially applied infrequently, mainly in patients with poorly localized surface EEG or evidence for bitemporal EEG abnormalities, who comprise 10% to 15% of surgical candidates. In the late 1960s and 1970s, some centers used intracranial EEG in all surgical candidates before resecting tissue for seizure control. This practice generated considerable historical and practical information but is no longer necessary. Now, some form of intracranial EEG investigation is performed on approximately 15% to 45% of adult patients at most centers that do epilepsy surgery. The proportion of patients considered for intracranial investigation depends on the referral population. In the pediatric population, both the need for and the practicality of intracranial recording diminish with decreasing age. Although invasive recordings can be done in infants,¹²⁷ they are rarely necessary.²³

Intracranial EEG is obtained with depth electrodes (Figs. 1 and 2), subdural strip electrodes (Figs. 3,4,5), subdural grid electrodes (Fig. 6), and/or epidural electrodes. They are multicontact, usually flexible electrodes that require an operative procedure for insertion in preselected locations based on prior evaluation. All of them can be used for chronic or acute recording over 1 to 4 weeks (rarely up to 8 weeks) for the purpose of localizing the epileptogenic zone by recording multiple spontaneous ictal events. They differ predominantly in the parts of brain that can be recorded. Only depth electrodes penetrate brain tissue directly. They are therefore best suited for recording from deep, buried structures such as amygdala and hippocampus (Figs. 1 and 2), planum temporalis, and insula. These electrodes are also well suited for recording from sulcal cortex, as well as near lesions in deep cerebral locations. Subdural electrodes are best suited for recording from the cortical surface, including dorsal, basal, and interhemispheric regions, and provide a method for safe, extensive cortical sampling. Subdural strip electrodes and depth electrodes can be inserted to record from multiple lobes and areas in both hemispheres. Subdural grid electrodes are restricted in insertion to previously defined zones.

The electrodes differ in the method of insertion, and with these differences come predictable differences in safety profiles. Depth electrodes are inserted with stereotactic MRI or computed tomographic (CT) guidance such that the trajectory, termination point, and location of each contact can be accurately predetermined.^62,87,91 These electrodes are usually inserted bilaterally, and accurate symmetric placements can be routinely achieved, not an insignificant accomplishment when comparison of the homotopic areas for onset and propagation of seizure activity is paramount in interpretation (see Depth Electrodes, Interpretation further on in this chapter). The strategy for selection of target locations of depth electrode placement differs somewhat in North American and certain European centers. The differences are partly a result of the more frequent combination of depth with additional implanted (subdural) electrodes in North America, thus allowing a more restricted use of depth electrodes for only some of the regions to be studied. The approach at some European centers developed from the stereo-electroencephalography method of Talairach and Bancaud, and aims to simultaneously and three dimensionally display all areas potentially involved in the onset, propagation, and clinical manifestations of the ictus, using only depth electrodes.^{11,12,25,47,64,103,104} Both are based on hypotheses developed from noninvasive study. The European approach may incorporate more regions of suspected propagation and evolution, a consideration that influences the surgical approach of those centers. Subdural strip electrodes are inserted with the least precision because they are passed through burr holes and “slid” in the subdural space to the desired target (Figs. 3,4,5). Adhesions caused by prior surgery, infection, or injury often deflect the desired placement, and thus strict precision and homotopic sampling are rarely achieved. Depth and subdural electrodes can be removed at the bedside. Subdural grid electrodes require craniotomy for insertion and removal and can be placed accurately. Subdural grid electrodes are usually placed only when a reasonably well-defined epileptogenic zone is known, and they serve to aid in precisely defining the limits of resection. They provide optimal cortical coverage for functional mapping with stimulation if seizures begin near language, motor, or sensory cortex.

Increasingly, the intracranial EEG methods are combined to achieve optimal sampling of all suspicious areas for seizure onset. They can all be used at any age, although the range of indications tends to be more restricted in young children, especially infants. The specific indications, construction, insertion, interpretation, limitations, risks, cost, and accuracy of each of these methods are the subjects of the remainder of this chapter.

Intracranial electrodes are indicated when it is reasonable to believe that a therapeutic resective procedure could be performed but the noninvasively obtained data are inadequate and more information is needed. Intracranial electrodes are then used to precisely identify the location of the ictal onset zone and can also delineate functionally important tissue that should be spared at the time of resection. Some protocols, particularly in certain European centers, also strive to delineate other regions of initial or early involvement that can be incorporated in designing the resection. Exact criteria for what constitutes sufficient concordant or discordant data vary somewhat from one epilepsy center to another.

Specific indications for intracranial electrodes in the evaluation of patients with refractory epilepsy include the following:

Although there are no set criteria that can be systematically used to decide whether intracranial recordings should be performed, these statements give broad guidelines. Factors such as the presence of a lesion and the type of epilepsy (temporal versus extratemporal) influence to various degrees the need for invasive recordings. In the great majority of cryptogenic cases, for instance, the use of intracerebral recordings remains essential.⁶⁵ In temporal lobe epilepsy, although diagnosis of mesial temporal lobe seizures and the side of ictal onset is now possible in the majority of patients using noninvasive investigation, additional long-term monitoring with intracranial electrodes is appropriate when the side of mesial temporal ictal onset is unclear, if neocortical or extratemporal ictal onset is considered possible, or when electroclinical arguments suggest that the epileptogenic zone extends outside the boundaries of a more traditional approach to the mesial temporal lobe resection. The delineation of extratemporal epileptogenic networks is much more difficult than that of temporal lobe regions of seizure origin, so that many more patients with extratemporal epilepsy undergo invasive EEG recordings: There are no clear extratemporal epilepsy “syndromes,” there is no equivalent of hippocampal sclerosis on MRI, scalp EEG is often mislocalizing and even mislateralizing, and there are no standardized resection techniques.

Certain electrode types are more suited to some locations than others and are therefore indicated under different circumstances. Because depth electrodes penetrate the substance of the brain, they are indicated when EEG recording is needed from buried cortex (e.g., hippocampus, amygdala, or basal or interhemispheric cortex inaccessible to subdural electrodes). In practice, depth electrodes are often used to record from the mesial temporal lobes when mesial temporal lobe epilepsy is suspected. They are also useful for recording from other deep cortical structures that can be involved (insula, planum temporalis). They are also essential for targeting deep regions of suspected seizure onset, for example, one or several heterotopic regions of gray matter.^2,76,105 Subdural electrodes overlie the surface of the brain and are most suited for bilateral neocortical EEG recording, often in combination with depth electrodes. Subdural grids are ideal for functional mapping of cortex, though regional preoperative localization is needed to select the (restricted) sites for placement. In practice, subdural strips are also often used additionally with grids for better sampling of other lobes ipsilaterally or contralaterally.