Temporal lobe seizures are the most commonly evaluated for surgery. Onsets of seizures from the temporal lobe can include epigastric, olfactory, and gustatory sensations, emotional changes, sense of familiarity or strangeness, hallucinations, staring, and automatisms, among others. One review [4] concluded that with temporal lobe epilepsy, abdominal aura had a 52% sensitivity and 90% specificity for localizing seizures to the temporal lobe. Seizures arising from temporal neocortex can have similar symptoms. For example, basal but not mesial temporal seizures can present with behavioral arrest or motor changes. Ictal theta activity was found to have an 85% probability for temporal lobe epilepsy and was 80–94% correct with respect to the side of seizure onset. Lateralized interictal spikes and possibly contralateral hand dystonia also were helpful. The authors thought that some of these also might help differentiate mesial from lateral temporal lobe epilepsy.

Febrile seizures are thought to have a relationship with mesial temporal sclerosis, as is found with mesial temporal lobe epilepsy; one report [5] found that only 2/21 patients with neocortical temporal lobe epilepsy had a history of febrile seizures. Seizure-free intervals were found to be less common with neocortical temporal lobe epilepsy than with mesial onset temporal lobe epilepsy. Despite neocortical onset, there nonetheless could be mild hippocampal atrophy. Patients could have tumors or heterotopias. About half could have decreased memory function on the Wada test. Independent contralateral spikes were rare. Some patients had experiential auras or motionless stares.

Frontal lobe epilepsy symptoms vary with the site of seizure onset [4]. With superior or interhemispheric onsets, there can be contralateral eye, head, or body turning with tonic or dystonic posturing. Orbital frontal seizures can include unusual behaviors including hypermotor activity such as rapid leg kicking or bicycling and can have autonomic findings, behavior arrest, and automatisms of other types. These characteristically occur frequently during sleep and last a relatively short period of time. Seizures from the frontal operculum can include salivation and swallowing. Inferior frontal onset can include findings referable to the face or to speech. Dorsolateral or dorsomedial onset seizures can include contralateral motor findings, premotor area seizures tonic version, and supplementary motor area seizures speech arrest, fencing postures, bilateral motor findings, and head version. It is important to note that frontal lobe regions can produce seizures that are similar to one another.

Seizures from the insula can include visceral, gustatory, and somatosensory symptoms, including laryngeal constriction or paresthesias [4]. Parietal lobe seizures can begin with somatosensory phenomena, and occipital lobe seizures can begin with visual auras and phenomena. However, both parietal and occipital lobe seizures can be locally silent, with symptoms related to the area of projection. For example, parietal lobe seizures can imitate superior frontal lobe seizures or can have sensorimotor symptoms.

Neuropsychological evaluation is important both in assessing baseline functioning and in determining whether there are aspects of function, which are below expectations. At times, these functions can be localized to specific regions of the brain, which in turn might be the sites of origin of the patient’s seizures.

The intracarotid sodium amobarbital or Wada test is performed less frequently now than had been the case in the past. When used, it has two purposes. With the test, a medication, which traditionally had been amobarbital, but now can be another such as midazolam, is injected so as to “anesthetize” one hemisphere for a few minutes while the other is tested. One looks for language function during the period of “anesthesia,” to see whether speech remains while the hemisphere is not functioning, and one presents items for the patient to remember. One also tests recall memory after the effects of the medication wear off, to see whether new memories could be encoded during the period of hemisphere inactivation. The idea is that if a function is intact during the period of drug-induced inactivation, the tested function is likely to be supported by the non-inactivated hemisphere.

Imaging is increasingly important in evaluating patients with intractable seizures, with magnetic resonance imaging (MRI) being the most important; one should always be performed if possible. Important findings include evidence of mesial temporal sclerosis or other abnormality, as well as evidence of tumor, dysplasia, vascular anomaly, developmental defects, or other changes. For patients with temporal lobe epilepsy, it is important to keep in mind that there can be bilateral atrophy on MRI in some patients, perhaps 20% [4]. Sometimes, surgery can nonetheless be performed on one side, if seizures only originate on that side, but it adds a consideration before deciding whether to operate and a consideration when counseling the patient with respect to possible postoperative memory problems. Neuroimaging [4] can show amygdala abnormalities in 55% of patients and changes in the enterorhinal cortex in 25% and in the fornix in 86%. In one study of patients with temporal lobe epilepsy and tumors [6], astrocytomas were found in 46%, gangliogliomas in 21%, oligodendrogliomas in 18%, dysembryoplastic neuroepithelial tumors in 6%, anaplastic astrocytomas in 6%, and meningiomas in 3%.

Dual pathology can occur in 15–52% of patients with hippocampal sclerosis, with etiologies including heterotopias, cortical dysplasia, and tumors. Vascular lesions including cavernous malformations and arteriovenous malformations can occur in about 5% of patients [4]. Although the term has often been used to describe the combination of hippocampal sclerosis plus another lesion, it also is used to describe the occurrence of two potentially epileptogenic lesions regardless of type.

Causes of extra temporal seizures in a series of 133 consecutive cases included [7] cortical dysplasia in 38% and tumor in 28%. They reported that 10/50 patients with cortical dysplasia also had tumors; 11/50 had infarcts or remote ischemic lesions. They found four with arteriovenous malformations, 3 with Sturge–Weber malformations, and 2 with Rasmussen’s encephalitis. 17% had no significant findings.

Positron-emission tomography (PET) scans can point to areas of decreased metabolic function which in turn can be area of epileptogenesis. Single-photon emission computed tomography (SPECT) studies can point to areas of altered function in a similar way with similar inferences regarding whether these might indicate where seizures are originating. Magnetic resonance spectroscopy (MRS) can point to areas with altered chemistry. fMRI is being developed as a possible alternative to the Wada test, using it to localize language, which has been relatively successful, as well as memory, which has not been as successful thus far.

Often noninvasive evaluation is sufficient to determine how and where to operate, but some patients need implanted electrodes as well. Depth and subdural electrodes are the ones principally used. Depth electrodes are thin “tubes,” each usually containing several electrodes and electrode wires, and which are directed through the skull and through the outer cerebral tissues, aiming at more medial locations such as mesial temporal lobes. However, there are electrodes along the tube so that more lateral locations including neocortex are recorded at the same time. Subdural electrodes are flat disks, usually a few millimeters in diameter, imbedded in Silastic or other plastics and placed over and around areas of interest. Both depth and subdural electrodes are used to localize the area of seizure onset. Subdural electrodes are commonly, and depth electrodes less commonly, stimulated electrically to determine the relationship of the area of seizure onset to regions controlling important functions such as movement, sensation, and language.

Complications of depth electrodes include asymptomatic subdural bleeding gliosis, degeneration, and microabscesses along electrode tract [8–10]. The incidence of bleeding or infection is between 0.5 and 5%. There can be a 25% overall rate of complications with subdural electrodes [11], including 12% infection, 11% transient neurological deficits, 2.5% epidural hematoma, 2.5% increased intracranial pressure, 1.5% infarction, and 0.5% death. Cerebrospinal fluid leakage also was common. The authors found that complications were more likely if there were more than 60 electrodes and if the grid was left in more than 10 days. Other risks included older patients, left-sided placement, and additional burr holes. They observed that complication risk likely was less now with improved technique.

Functional localization can be performed in one of two ways. One can alter the brain, for example with cortical stimulation, and assess the behaviors that occurred during the alteration. An example would be to see whether there is hand or other movement during stimulation. One also can alter behavior and then assess the brain during the behavior. An example might be asking the patient to begin to read and then seeing whether there is reading arrest during stimulation. Cortical stimulation is generally performed with recurrent pulses. These should be alternating in polarity, so that the resulting stimulation is charge-balanced, to avoid complications due to metal deposit on an electrode. We [12] have used 0.3-ms duration alternating polarity square wave pulses, delivered at 50 pulses per second, with stimulation duration varying but generally 1–2 s initially and then up to about 5 s for language testing. Intensities that are needed to obtain stimulation-induced changes vary; with the device we use, they can go up to 17.5 mA. It is important to emphasize that the reliability of results can depend on the intensity of stimulation. If you stimulate at too low an intensity, you can get a false-negative result. If you stimulate at too high an intensity, you can get afterdischarges which can produce false positives because of the spread of the afterdischarges and also can cause seizures. One should begin at a low intensity, 0.5–1 mA, and increase in increments of 0.5–1 mA. Keep in mind that the above is in milliamps, but the important parameter is charge density, which depends on not only current but also electrode surface area.

It is also important to emphasize that stimulation only assesses the cortex directly under the stimulated electrodes. Charge density drops relatively rapidly with increased distance from the actual location of the electrodes. Also, 7/8 of the current is shunting through the cerebrospinal fluid [13, 14].