CHAPTER 50 Epilepsy
Seizures develop when cortical neurons undergo abnormal, excessive, hypersynchronous electrical discharges due to impaired regulation of neuronal excitation and inhibition. Epilepsy is a condition characterized by recurrent spontaneous seizures. It is a common disorder, present in 0.4% to 1% of the population.1
Seizures can be categorized as either partial or generalized. Partial seizures originate from a localized region of the brain. Generalized seizures involve both cerebral hemispheres simultaneously. Partial seizures are subdivided into complex partial seizures, with loss of consciousness, and simple partial seizures, without loss of consciousness. Partial seizures may generalize secondarily by spread from one area to another.
Seizure classification has therapeutic and prognostic implications for the care of the epilepsy patient. Some seizure classifications are clinical, like that used by the International League against Epilepsy (ILAE). These include findings on electroencephalography (EEG), patient age, and clinical seizure type to help categorize seizures. Other classifications are anatomic. The focus in this chapter is to address the anatomic substrates of seizures and epilepsy that are identifiable by neuroimaging. Clinical features and classifications are not discussed in detail.
Generalized epilepsy is usually well controlled by medication. Epilepsy associated with partial complex seizures is often medically intractable and may require surgery to achieve seizure control. Overall, 15% to 30% of adult patients with partial seizures are not well controlled with antiepileptic medication.1 A major goal of epilepsy imaging is to identify an epileptogenic substrate or lesion that is amenable to surgery or other directed treatment. A second goal is to identify any syndromic cause of epilepsy.
MRI is the main modality of choice for evaluating patients with epilepsy, owing to its high soft tissue contrast, multiplanar imaging capability, lack of ionizing radiation, and higher sensitivity compared with CT. CT is useful in the initial evaluation of seizures, especially in the setting of trauma, acute focal neurologic signs, or fever. Published guidelines recommend that nonemergent MRI should be performed in most patients with epilepsy. Patients with febrile seizures and those with primary idiopathic generalized epilepsy do not need imaging unless there are complicating factors.
The sensitivity of MRI for detecting seizure foci is dependent on the population investigated. The sensitivity is relatively low for those with new onset of seizures but high for patients with medically refractory partial epilepsy. In one series of 300 consecutive patients with new-onset seizure, an anatomic substrate was identified by MRI in 14% (38/263).2 In that study, all of the patients with an imaging abnormality had partial seizures. Patients with primary generalized seizures did not have imaging abnormalities. In another series of children with a new diagnosis of epilepsy, 13% (62/388) had a structural abnormality demonstrated by MRI.3 Conversely, in patients with medically intractable epilepsy, the sensitivity of MRI has been reported as 82% to 86%.4 In a series of 117 patients with refractory partial seizures who had surgery, the sensitivity for detecting anatomic substrates of seizure was 32% (35/109) for CT and 95% (104/109) for MRI.4
In surgical candidates, MRI is essential for identifying a structural lesion, localizing it, and demonstrating its relationship to eloquent regions of the brain. The MRI findings must be correlated with electrophysiologic and clinical data to avoid falsepositive localization of the epileptogenic substrate.5 When MRI findings and noninvasive electrophysiologic data are concordant, surgery may be undertaken without resorting to invasive electroencephalography. MRI also helps to prognosticate the potential for successful surgical control of seizures by identifying and characterizing the seizure substrate.6 Postoperative MRI helps to localize electrode placement, identify surgical complications, and elucidate causes for treatment failure, such as recurrent or residual structural lesion.
Complex partial seizures may arise from any portion of the cerebral cortex. The specific clinical manifestations depend on the region involved. By definition, the epileptogenic region or zone is the portion of tissue that, when removed, will result in a seizure-free state. Imaging workup is directed toward identifying the epileptogenic region. No single test perfectly defines the epileptogenic region, so clinical assessment, neuroimaging, EEG, and neuropsychological tests are used in conjunction to infer the localization and extent of the epileptogenic region. When all tests are concordant, there is a high likelihood that the epileptogenic region or a portion of it has been identified. CT, MRI, and pathologic evaluation can also identify the epileptogenic lesion, the anatomic substrate thought to be responsible for the epilepsy. The epileptogenic zone encompasses at least part of the epileptogenic lesion.
A few clinical terms are commonly used in epilepsy. Seizure semiology is the study and description of seizure signs and symptoms. An aura is a subjective or sensory simple partial seizure that may precede or herald a more severe complex partial or generalized seizure. An automatism is an intraictal repetitive motor activity often occurring during cognitive impairment, such as lip smacking. After a seizure, patients may experience transitory functional neurologic abnormalities referred to as postictal phenomena. A postictal unilateral neurologic deficit is referred to as a lateralizing phenomenon. A postictal phenomenon that is nonfocal and involves impaired cognition, psychosis, or amnesia is referred to as a nonlateralizing postictal phenomenon. Postictal hemiparesis is a common lateralizing sign called Todd’s paralysis.
Temporal lobe seizures can begin in the temporal lobe or generalize to the temporal lobe after originating elsewhere. The temporal lobe is the most common origin for complex partial seizures, both motor and sensory. A common temporal lobe motor seizure is dystonia of the contralateral upper extremity. Commonly reported temporal auras include visceral sensations, déj vu, gustatory sensations, olfactory sensations, and staring. Automatisms may be present, often related to oroalimentary function, speech, facial expressions, or laughing. Postictal aphasia or amnesia can be seen.
Occipital seizures are characterized by visual and ocular symptoms, often with visual auras. Parietal seizures may include visual hallucinations, body movement sensations, and vertigo, but most do not have an identifiable aura. Frontal lobe seizures are poorly understood and have a wide variety of clinical presentations. They include unusual behavioral, sensory, autonomic, and hallucinatory patterns. Frontal motor seizures often result from involvement of the primary motor cortex. As opposed to temporal lobe seizures, which often have a long duration, frontal seizures may last only a few seconds. This brief period may be too short to permit adequate ictal SPECT injections (see later).
Many factors determine which imaging protocols are “appropriate” or “optimal” for evaluating seizure disorders. These include patient age, the specific features of the population being imaged, and the imaging characteristics of the epileptogenic substrates expected at that age in that population. The imaging characteristics of normal brain change with age, so the optimal imaging protocol must also change with age. This particularly applies to the incompletely myelinated infant brain. The likely causes of epilepsy also vary with age, so imaging protocols must be optimized for the lesions most likely at each age (Table 50-1).
In infants younger than 18 months of age, myelination is still incomplete. The water content of the brain is higher, so fluid-attenuated inversion recovery (FLAIR) sequences are less useful for detecting areas of abnormal T2-weighted (T2W) signal. The interface between gray and white matter is usually indistinct on spoiled gradient-recalled-echo (SPGR) sequences, especially when the patient is younger than 6 months of age, so inversion recovery and fast spin-echo (FSE)/T2W sequences may be needed to optimize gray matter/white matter differentiation in these patients (Fig. 50-1). In neonates, infection, stroke and malformations of cortical development (MCD) are primary considerations, so attention is directed toward display of the cortex. Mesial temporal sclerosis is not a diagnostic consideration in this age group, so coronal oblique imaging through the hippocampus is not employed routinely. Diffusion-weighted imaging (DWI) and administration of gadolinium (Gd) may be useful in these patients, particularly in patients with new-onset seizures.
FIGURE 50-1 Immature myelination. A, Coronal spoiled gradient (SPGR) MR image obtained from a 4-month-old infant appears to demonstrate diffuse cortical thickening. There is poor differentiation between gray and white matter. This appearance, however, is due to the immature myelination, demonstrating the limited value of SPGR sequences before 12 months of age. B, Coronal FLAIR MR image from the same patient shows diffuse hyperintensity of the incompletely myelinated white matter. The utility of this sequence is also limited before 12 to 18 months of age. C, Coronal T2W FSE MR image from the same patient shows better gray matter/white matter differentiation.
In patients older than age 50 years, stroke and neoplasm are more common causes of new-onset seizure. Therefore, DWI is required to detect areas of ischemia; a contrast agent should be administered routinely to help detect neoplasms (Fig. 50-2). The frequency of hippocampal sclerosis is low in this population, so sequences tailored for evaluation of the hippocampus are not used routinely.
FIGURE 50-2 Right temporal lobe glioblastoma. A, Coronal FLAIR MR image of a 65-year-old patient who presented to the emergency department with new-onset seizures demonstrates a large medial right temporal lobe mass with surrounding vasogenic edema. B, Coronal T1W MR image post contrast agent administration shows thick irregular ring enhancement of the temporal lobe lesion. Pathology after resection confirmed this was a glioblastoma multiforme.
The patients most likely to be evaluated for medically refractory partial epilepsy are those between the ages of 18 months and 50 years. This population will have the highest yield for detection of structural pathology by MRI, including hippocampal sclerosis. These patients have completed myelination and can undergo detailed evaluation for subtle epileptogenic abnormalities. In this age group, special protocols are needed to increase the likelihood of identifying hippocampal sclerosis and cortical malformations.7 Administration of a contrast agent is not needed routinely unless there is specific concern for tumor, vascular malformation, or infection. One notable exception to this is in the patient with hemiatrophy, where use of contrast enhancement may reveal the Sturge-Weber malformation.
The coronal plane oriented perpendicular to the long axis of the hippocampus is ideal for assessing signal change within the hippocampus, abnormal hippocampal architecture, and hippocampal atrophy (Fig. 50-3).
FIGURE 50-3 Normal coronal temporal lobe anatomy. Coronal inversion recovery MR image of normal temporal lobe anatomy. Large arrow, hippocampus; BP, brachium pontis (middle cerebellar peduncle); PHG, parahippocampal gyrus; TS, temporal stem; FG, fusiform gyrus; ITG, inferior temporal gyrus; MTG, middle temporal gyrus; STG, superior temporal gyrus; small arrow, tail of the caudate nucleus.
FLAIR pulse sequences are most sensitive for detecting signal abnormalities of the hippocampus. However, it is best to use a combination of coronal FSE/T2W sequences and coronal FLAIR for assessing the hippocampus, because the normal hippocampus may appear slightly hyperintense as compared with neocortical gray matter on FLAIR sequences.8
T1-weighted (T1W) gradient volume acquisitions (SPGR or MP-RAGE) are excellent for evaluating the morphology of the hippocampus. The raw data from these images can be reformatted into any plane that is helpful for qualitative and volumetric analysis of the hippocampus.
Spectroscopy and T2 relaxometry may be used for additional evaluation of the hippocampus and medial temporal lobes. T2 relaxometry obtained from multi-echo T2W sequences is especially useful in cases in which the findings on visual analysis are equivocal or for lateralizing the seizure focus when abnormalities are evident in both hippocampi.
Cortical malformations may be difficult to detect without high-resolution techniques and pulse sequences that optimize gray matter/white matter differentiation. T1W gradient 3D volume sequences (SPGR or MP-RAGE) with thin slices can provide high spatial resolution with superb distinction between gray and white matter. Inversion recovery sequences can provide excellent corticomedullary contrast for detecting subtle thickening of the gray matter and indistinctness of the gray matter/white matter junction. FLAIR and T2W sequences are recommended to reveal any hyperintensity in the subcortical and deep white matter indicative of congenital malformation, as well as for detecting hyperintense cortical neoplastic, inflammatory, or gliotic abnormalities.
High field magnets (3 T or greater) and high-resolution phased-array coils help to detect cortical malformations. Photographic image reversal of inversion recovery or T2W sequences may be helpful. Multiplanar or 3D reconstructions of volume-acquired data can demonstrate sulcal morphologic abnormalities, cortical dysplasia, indistinctness of the gray matter/white matter interface, and the relationship of developmental abnormalities to eloquent areas of the brain. However, the location and extent of cortical dysplasia identified by MRI may not correlate directly with the seizure semiology or the electrophysiologic data.5
MRI postprocessing techniques have advanced the quantitative and qualitative evaluation of epileptogenic abnormalities. Computerized segmentation of gray and white matter can detect developmental anomalies in epilepsy patients. Other image processing advancements such as curvilinear reformatting, texture analysis, and automated quantitative methods may be helpful in individual cases.
The term mesial temporal sclerosis signifies scarring and volume loss of medial temporal structures: the hippocampus, the amygdala, and the parahippocampal gyrus (including the entorhinal cortex). Alternate names include hippocampal sclerosis and Ammon’s horn sclerosis.
Epilepsy affects 0.5% to 1% of the population. Up to 30% of cases are medically intractable or medically refractory, that is, seizures persist despite optimal medical therapy.1 Temporal lobe epilepsy accounts for about 70% of intractable epilepsy. In patients undergoing surgery for intractable epilepsy, hippocampal sclerosis is found pathologically in 60% to 70%. MRI has also shown hippocampal sclerosis in patients with medically controlled, complex partial epilepsy and in relatives of epileptics without clinical seizures.
Patients with hippocampal sclerosis often have a history of a childhood insult, usually complicated febrile seizure or encephalitis before the age of 5. After a quiescent period, recurrent temporal lobe seizures begin during the second decade. A subset of patients with medial temporal lobe epilepsy have paradoxical medial temporal lobe epilepsy, in which the MRI is normal. In this group, hippocampal gliosis occurs without neuronal loss. The postoperative outcome is somewhat poorer in this group than in classic hippocampal sclerosis.9
Dual pathology is the term to describe the coexistence of hippocampal sclerosis with another epileptogenic substrate. This occurs in 8% to 22% of surgical epilepsy patients.10 The most frequent coexisting epileptogenic substrate is cortical dysgenesis. Dual pathology is associated with less favorable surgical outcome, unless both the hippocampus and the other substrate are resected. In this situation, a residual sclerotic hippocampus or a residual extrahippocampal substrate will become the epileptogenic source after the initial surgery.
The complex pathophysiology leading to hippocampal sclerosis is incompletely understood. Multiple factors may cause this condition. The “two-hit” hypothesis proposes that an initial precipitating injury (e.g., complicated febrile seizures or encephalitis) must occur in the presence of a predisposing factor (e.g., genetic disposition or developmental anomaly) that increases vulnerability. The presence of dual pathologic processes within the temporal lobe supports this proposal. Hippocampal sclerosis is associated with a second abnormality on MRI in 15% to 30% of cases.10 “Kindling,” induction of a secondary seizure focus by repeated exposure from a primary seizure focus, may also contribute to the development of, or bilateral extension of, hippocampal sclerosis.
The hippocampus is a curved structure situated along the medial aspect of the temporal lobe. It can be divided into three regions by its morphology and relationship to the midbrain. The anterior, expanded region is called the hippocampal head or pes hippocampus. Three or four longitudinal striations groove the superior surface of the head, giving it the shape of fingers (the digitations of the pes hippocampi). The midregion is the cylindrically shaped hippocampal body, which lies adjacent to the midbrain. The posterior hippocampal tail rapidly narrows behind the brain stem.
In coronal sections, the hippocampus appears as two interlocking U-shaped layers of gray matter, the dentate gyrus and cornu ammonis. The cornu ammonis is divided into four segments: CA1, CA2, CA3, and CA4. CA1 is called the Sommer sector and is especially vulnerable to hypoxia. CA4 lies within the curve of the dentate gyrus and is also called the end folium. The cornu ammonis extends laterally to merge with the subiculum, which extends laterally, in turn, to merge into the neocortex of the parahippocampal gyrus. The convex ventricular surface of the hippocampus and dentate gyrus is covered with ependyma. A thin lamina of white matter between the ependyma and the superior surface of the hippocampus is designated the alveus. This is a major tangential communication pathway. The alveus passes medially over the dentate gyrus to form a free margin designated the fimbria. Together the alveus and the fimbria constitute the fornix. The amygdala is a gray matter structure located superomedial to the tip of the temporal horn of the lateral ventricle. The uncal recess of the temporal horn separates the amygdala from the hippocampal head. When the uncal recess is not visible, the alveus (i.e., the thin layer of white matter on the superior surface of the hippocampus) may help to separate the amygdala from the hippocampus.
Mesial temporal sclerosis affects variable portions of the temporal lobe and may be unilateral, bilaterally asymmetric, or bilaterally symmetric. Ten to 20 percent of cases of hippocampal sclerosis are bilateral. Bilaterality is often associated with concurrent developmental anomalies. The sclerosis may affect the anterior hippocampus (head) most severely or extend the entire anteroposterior length of the hippocampus. Rarely, the amygdala is affected predominantly. The entire temporal lobe may be affected.
In hippocampal sclerosis there is usually loss of more than 50% of the neurons, notably loss of pyramidal cells in regions CA1, CA2, and CA4 and loss of granule cells within the dentate gyrus (Fig. 50-4). Hippocampal reorganization and changes in energy metabolism are also seen in hippocampal sclerosis. Findings of reorganization include abnormal axonal sprouting and loss of interneurons, which is thought to change the balance of neuronal excitation and inhibition.
FIGURE 50-4 Hippocampal sclerosis histology. A, Coronal histologic section of an adult hippocampus with hippocampal sclerosis. There is significant loss of pyramidal cells within the CA1 and CA4 fields. T, temporal lobe; F, fimbria; D, dentate gyrus; M, molecular strata of the dentate gyrus. B, Coronal histologic section of a normal adult hippocampus. S, subiculum; 1 to 4, CA1, CA2, CA3, and CA4 fields of the cornu ammonis, respectively; small arrowheads, dentate gyrus; M, molecular strata of the dentate gyrus; F, fimbria; A, alveus; large arrowhead, minimal residual hippocampal sulcus; T, temporal horn of the lateral ventricle. C, High-resolution thin-section coronal T2W MR image of a normal adult hippocampus. The general aspects of the hippocampal internal architecture are visible. S, subiculum; 1 and 2, CA1 and CA2 fields of the cornu ammonis; A, alveus; F, fimbria; T, temporal horn of the lateral ventricle. D, High-resolution thin-section coronal SPGR MR image of the same patient. S, subiculum; 1 and 2, CA1 and CA2 fields of the cornu ammonis; A, alveus; F, fimbria; T, temporal horn of the lateral ventricle.
(A, modified from Bronen RA, Cheung G, Charles JT, et al. Imaging findings in hippocampal sclerosis: correlation with pathology. AJNR Am J Neuroradiol 12:933-940, 1991, © by American Society of Neuroradiology; B from Kier EL, Kim JH, Fulbright RK, et al. Embryology of the human fetal hippocampus: MR imaging, anatomy, and histology. AJNR Am J Neuroradiol 18:530, 1997, © by American Society of Neuroradiology.)
CT does not display hippocampal sclerosis directly. In severe cases, it may demonstrate gross hippocampal atrophy, with enlargement of the adjacent temporal horn, small size of the hippocampal formation, and reduced size of the ipsilateral fornix.
More than 75% of patients with surgically proven hippocampal sclerosis show prolonged T2 relaxation time (hyperintense signal on T2W images), hippocampal atrophy, and loss of the internal architecture of the hippocampus (Fig. 50-5). Other MRI findings associated with hippocampal sclerosis include loss of the digitations along the hippocampal head, dilatation of the adjacent temporal horn, atrophy of the white matter in the adjacent parahippocampal gyrus, increased T2 signal in the anterior temporal white matter, and secondary atrophy of the fornix and mammillary bodies due to degeneration of hippocampal tracts (Fig. 50-6). Overall, qualitative (i.e., visual) assessment of MRI detects hippocampal sclerosis with a sensitivity in the range of 75% to 90%.
FIGURE 50-5 Normal versus abnormal hippocampus. A, Coronal SPGR MR image from a patient with long-standing seizures shows a normal-sized right hippocampus (white arrow) in comparison with the smaller, atrophic left hippocampus (black arrow). B, Coronal T2W MR image of the same patient again shows the atrophy of the left hippocampus (black arrow), but without a significant signal difference from that of the right hippocampus (white arrow). C, Coronal FLAIR MR image demonstrates normal signal within the right hippocampus (white arrow) and also definitely depicts slight T2 prolongation within the atrophic left hippocampus (black arrow), suggesting mesial temporal sclerosis.
FIGURE 50-6 Associated findings with hippocampal sclerosis. A, Coronal T2W MR image in a patient with left hippocampal sclerosis shows asymmetric thinning and atrophy of the left fornix (black arrow) compared with the normal right fornix (white arrow). B, A more anterior coronal T2W MR image of the same patient also shows atrophy of the left mammillary body (black arrow) as compared with the normal right mammillary body (white arrow).
Quantitative MR methods, such as hippocampal volumetry and T2 relaxometry, increase the sensitivity for detecting hippocampal sclerosis to 90% to 95%. These quantitative methods are especially useful when hippocampal sclerosis occurs bilaterally without obvious T2 signal changes (Fig. 50-7). Proton MR spectroscopy may also be used quantitatively.
The criteria set forth by Watson and coworkers11 are best for defining the anatomic boundaries needed to measure the volumes of the hippocampus and amygdala. Each center needs to establish its own normative data for hippocampal volume because the precise paradigm used in the quantitative imaging program varies from center to center. One needs to normalize for such variables as head size, patient age, patient gender, and the hemisphere imaged (right or left).
Quantitative measurements of T2 signal (T2 relaxometry) are obtained on a single-slice multi-echo sequence through the hippocampal body. The T2 signal is measurably elevated in approximately 70% of cases with hippocampal sclerosis.12 Measurements of T2 relaxometry are particularly useful in detecting abnormalities of the amygdala (amygdala sclerosis), which may be difficult to identify on routine anatomic imaging.13 Concurrent involvement of the amygdala in mesial temporal sclerosis reduces the likelihood that resective surgery will achieve a seizure-free outcome from 80% to 50%.14
Proton magnetic resonance spectroscopy (MRS) can be useful in evaluating the hippocampus and lateralizing a seizure focus. MRS of the hippocampus is typically performed using single voxel spectroscopy, which often includes some of the adjacent mesial temporal structures. The contralateral hippocampus is interrogated for comparison. The most important spectroscopic finding in mesial temporal sclerosis is decreased N-acetyl-aspartate (NAA). Reduced ratios of NAA to creatine or NAA to (choline + creatine) reflect metabolic dysfunction and/or neuronal loss. NAA may be a dynamic marker of neuronal dysfunction, rather than a simple sign of decreased numbers of neurons. Abnormally low NAA concentrations have been reported to recover in the unoperated temporal lobe after successful contralateral temporal lobectomy.15 In patients with bilateral temporal lobe MRI abnormalities, reduced NAA ratios successfully lateralize the site of epileptogenesis in 65% to 96% of cases.16 In temporal epilepsy patients without MRI abnormalities, reduced NAA ratios have correctly lateralized the seizure focus in at least 20% of cases. Bilaterally reduced NAA to creatine ratios have been associated with failed resective surgery.
Quantitative methods have great value for research, allowing the investigator to test hypotheses correlating MRI data with clinical and pathologic data. Hippocampal volume has been correlated with cell loss, frequency of childhood febrile seizures, memory functions, duration of epilepsy, and successful surgical outcome.17,18 One study correlated recurrent temporal lobe seizures with hippocampal volume loss, whereas generalized seizures were linked to progressive neuronal damage.15 This study supports early intervention for seizure control to prevent progressive brain damage. Although quantitative MRI has many advantages for research, routine clinical use of quantitative measures faces some obstacles: operator time, the need for dedicated personnel, workstation, and software, and the requirement of a truly representative data sample of normal controls.
Several neurotransmitters have been evaluated using proton spectroscopy, including glutamate, glutamine, and γ-aminobutyric acid (GABA). Future investigations may prove these techniques useful for elucidating neurochemical derangements in epilepsy syndromes and pharmacokinetic monitoring of medical therapy.
Sclerotic hippocampi may show abnormally elevated ADCs,19 even when conventional MRI is normal, likely reflecting early change. The ADCs of the abnormal hippocampi are higher than the ADCs of the corresponding contralateral hippocampi and higher than the ADCs from hippocampi of healthy volunteers. Interestingly, in patients with one sclerotic hippocampus, the ADCs of the contralateral normal-appearing hippocampi are higher than the ADCs of healthy volunteers. The ADC abnormalities in the contralateral normal-appearing hippocampi may resolve after resection of the ipsilateral abnormal hippocampus, suggesting that the biochemical change may be related to ongoing epileptic activity. In one study, ADC maps correctly lateralized the abnormality in 100% of patients. When the conventional MRI is nonlateralizing, however, the ADC values are also less helpful in lateralization.
DTI has shown reduced fractional anisotropy and increased mean diffusivity in the hippocampi on the side of seizure lateralization (Fig. 50-8) versus the contralateral normal-appearing hippocampus in control subjects.20 Compared with normal controls, patients with hippocampal sclerosis and negative conventional MRI have higher mean diffusivity and low fractional anisotropy in the hippocampi bilaterally. Decreased fractional anisotropy values have also been demonstrated in the uncinate fasciculus on the side of seizure lateralization, suggesting that this tract has a role in the spread of temporal lobe seizures.
FIGURE 50-8 Right mesial temporal sclerosis. A, Coronal FLAIR MR image in a patient with chronic seizures shows classic findings of mesial temporal sclerosis with an atrophic, hyperintense right hippocampus (arrow) compared with the left hippocampus with normal size and signal. B, Axial SPGR MR image of the same patient also demonstrates the atrophy of the right hippocampus (arrow) when compared with the normal left hippocampus. C, Diffusion tensor imaging (average diffusivity map) demonstrates increased average diffusivity within the abnormal right hippocampus (arrow). D, Axial interictal 99mTc HMPAO SPECT image demonstrates hypoperfusion of the right hippocampus/medial temporal lobe (arrow). E, Axial PET image shows subtly decreased glucose metabolism within the right medial temporal lobe (arrow).
Diffusion tensor fiber-tracking of major white matter tracts can be particularly helpful in surgical planning. In conjunction with functional MRI, tractography can demonstrate connectivity between noncontiguous functionally eloquent cortical regions. Connectivity mapping may improve understanding of seizure origin and spread and lead to innovative minimally invasive surgical strategies.
In the planning of lesional resective surgery, many epilepsy centers use functional MRI (fMRI) to localize eloquent regions of the brain that should be preserved. Less frequently, functional mapping can be used to localize the activated cortex during an ictal seizure. fMRI uses blood oxygen level–dependent (BOLD) changes in T2* signal to demonstrate areas of increased neuronal activity and resultant increased perfusion. EEG and fMRI can be used in combination to obtain whole-brain maps of interictal epileptic spikes and their co-registered associated BOLD changes.
18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) is used to map cerebral glucose metabolism. This study is usually performed in an interictal period. Routine FDG-PET has a sensitivity of more than 70% for lateralization of mesial temporal epilepsy. The epileptogenic temporal lobe demonstrates interictal hypometabolism, that is, decreased uptake of the glucose analogue. Hypometabolism may also occur in the contralateral temporal lobe and in ipsilateral extratemporal locations such as the frontoparietal cortex, basal ganglia, and thalamus. These additional areas, however, all demonstrate a lesser degree of hypometabolism compared with the epileptogenic temporal lobe. Bilateral asymmetric temporal lobe hypometabolism is commonly seen in mesial temporal epilepsy. One should keep in mind that antiepileptic medications, particularly barbiturates, can result in a global decrease in glucose metabolism.
The area of PET abnormality may be significantly larger than that requiring surgical resection. PET and SPECT do not give information about the nature of the underlying substrate but can help to focus the MRI interpretation, especially in cases of subtle MR findings.
PET with 11C-flumazenil (FMZ) is a newer technique that has shown much promise. GABA is the most important inhibitory neurotransmitter. Flumazenil specifically binds to GABA receptors, which are decreased in mesial temporal sclerosis. Early studies have shown 11C-FMZ abnormalities to be better localized than 18F-FDG abnormalities,21 so this technique may prove particularly useful in cases in which MRI is negative.
Single-photon emission CT (SPECT) and subtraction ictal-interictal SPECT co-registered to MRI (SISCOM) have been used. In epileptic patients, SPECT may be used to evaluate brain perfusion and to map regional cerebral blood flow (rCBF). SPECT can be performed using a number of different radiopharmaceuticals. The study acquires data in three dimensions, after which cross-sectional images can be generated in any plane by using filtered back-projection. Interictal SPECT characteristically demonstrates hypoperfusion in the epileptogenic region but has low sensitivity and specificity. Ictal SPECT demonstrates hyperperfusion in the epileptogenic region and has a very high sensitivity. The injection can be performed ictally, during the actual seizure, because the tracer is localized very quickly after injection. The actual imaging (data acquisition) can then be delayed up to 4 hours postictally, because the tracer remains localized at the original uptake site for long periods. This makes ictal SPECT far more feasible than ictal PET. Achieving an ictal injection, however, can be logistically challenging and requires a dedicated setup. SISCOM provides better localization than either ictal SPECT alone or qualitative comparison of ictal to interictal SPECT. This technique may be especially useful for patients in whom MRI is either negative or demonstrates diffuse or multifocal abnormalities.