Neuroimaging in the evaluation of seizures and epilepsy

INTRODUCTION — The term "epileptic seizure" refers to a transient occurrence of signs and/or symptoms due to abnormally excessive neuronal activity of the cerebral cortex. Epilepsy is a condition characterized by recurrent, unprovoked seizures. The diagnosis of epilepsy is often not straightforward, and misdiagnosis is relatively common [1]. A detailed and reliable account of the event by an eyewitness is crucial to the diagnostic evaluation, but may not be available [2].

The purpose of the diagnostic evaluation in a patient with seizures is to provide evidence that helps confirm or refute the diagnosis of epilepsy and to identify the cause of epilepsy and/or to classify the epileptic syndrome. Neuroimaging also has a critical role in the evaluation of patients with refractory epilepsy for epilepsy surgery.

This topic will discuss neuroimaging in the diagnostic evaluation of adults with possible epileptic seizures. Other aspects of diagnostic testing for seizures and epilepsy are discussed separately. (See "Evaluation and management of the first seizure in adults" and "Seizures and epilepsy in children: Clinical and laboratory diagnosis" and "Electroencephalography (EEG) in the diagnosis of seizures and epilepsy" and "Video and ambulatory EEG monitoring in the diagnosis of seizures and epilepsy".)

COMPUTED TOMOGRAPHY — Computed tomography (CT) is commonly ordered in patients presenting with new-onset seizure to an emergency department. It is generally available quickly in that setting and is used to exclude acute neurologic problems that require urgent intervention [3]. CT is able to identify hemorrhages, gross structural malformations, large tumors, and calcified lesions. Skull-based CT may be helpful in identifying bone deficits through which temporal encephaloceles protrude. CT angiography may be useful in characterizing vascular malformations such as arteriovenous malformations or arteriovenous fistulas. (See "Evaluation and management of the first seizure in adults".)

However, in most nonemergency situations, magnetic resonance imaging (MRI) is more sensitive than CT and is the neuroimaging study of choice.

MAGNETIC RESONANCE IMAGING — Most patients suspected of having had an epileptic seizure should have a neuroimaging study. The purpose is to identify a structural etiology for epilepsy. The mainstay of elective neuroimaging is MRI, which has not only higher sensitivity than CT for most epileptogenic lesions but also better spatial resolution and soft-tissue contrast [4-6]. In addition, it allows imaging in multiple planes as well as functional cerebral assessment through different techniques.

Exceptions to the need for neuroimaging include children with simple febrile seizures and children whose clinical history and electroencephalography (EEG) are consistent with benign partial epilepsy of childhood or idiopathic generalized epilepsy. (See "Clinical features and evaluation of febrile seizures" and "Self-limited focal epilepsies of childhood" and "Epilepsy syndromes in children".)

Specific pathologies — Structural causes of epilepsy that can be identified by MRI include [7,8]:

●Mesial temporal sclerosis (hippocampal sclerosis) (image 1 and image 2)

●Malformations of cortical development (eg, cortical dysplasia) (table 1 and image 3 and image 4)

●Brain tumors

●Vascular malformations (image 5)

●Cerebral infarction, cerebral hemorrhage

●Traumatic brain injury

●Infections, including encephalitis, cerebral abscess, granulomas, and cysts (neurocysticercosis)

●Temporal encephaloceles

Mesial temporal sclerosis, also known as hippocampal sclerosis, is the most commonly diagnosed focal structural abnormality in patients with epilepsy. While most cases present in childhood, it is not uncommon for this disorder to first appear in young adults.

Primary imaging signs of mesial temporal lobe sclerosis (image 1 and image 2) include:

●Reduced size

●T2- and FLAIR-hyperintense signal

●Abnormal morphology (eg, loss of stratum lacunosum and moleculare)

Secondary signs include ipsilateral:

●Atrophy of the fornix, mammillary body, or both

●Enlargement of the temporal horn, choroidal fissure, or both

●Reduced volume of white matter in the parahippocampal gyrus

●Global reduction of temporal lobe white matter volume

Surgery or laser interstitial thermal treatment is often curative in patients who do not become seizure free on medication. (See "Focal epilepsy: Causes and clinical features", section on 'Hippocampal sclerosis' and "Surgical treatment of epilepsy in adults".)

Malformations of cortical development (eg, focal cortical dysplasia) (table 1) are the second most common structural etiology for epilepsy [9,10]. When a patient with focal-onset epilepsy is found to have a normal MRI, an undetected focal cortical dysplasia is typically considered to be the most likely underlying lesion.

With increasing use of higher-field MRI (ie, 3-Tesla and 7-Tesla), focal cortical dysplasia is being identified in a greater number of epilepsy patients. MRI is more sensitive for identifying focal cortical dysplasias with more severe pathologic grade than for milder lesions, and these are more amenable to surgical cure [11,12]. Subtle findings such as blurring of the grey-white junction or FLAIR-hyperintensity in subcortical white matter may be missed on initial interpretation of the MRI and identified when the study is re-reviewed (image 3) [10]. These lesions are congenital, and related epilepsy usually presents in childhood. However, it is not rare for epilepsy to first develop in young adults.

Low-grade glial or glioneuronal neoplasm is a cause of seizure in children and young adults (image 6). In older patients, cortical scarring (eg, encephalomalacia) after trauma or ischemic stroke, or higher-grade neoplasm may be a cause of seizure. (See "Seizures and epilepsy in older adults: Etiology, clinical presentation, and diagnosis".)

Certain infections, especially neurocysticercosis caused by Taenia solium, are common etiologies of epilepsy in endemic populations. When neurocysticercosis is suspected, MRI with contrast is useful for identifying cysts and evaluating disease activity (image 7). However, CT may add to the diagnostic evaluation, as CT is more sensitive than MRI for detecting small areas of calcification. (See "Cysticercosis: Clinical manifestations and diagnosis" and "Cysticercosis: Epidemiology, transmission, and prevention".)

Sensitivity — Most individuals with new-onset epilepsy will not have a structural lesion on MRI; in one case series, the yield was 14 percent [7]. Among reported studies, a wide range (1 to 57 percent) of neuroimaging studies in patients with epilepsy are abnormal [4,13-16]. These differences reflect the technology used (eg, CT versus MRI, and low- or high-field MRI), the patient population studied, and the prevalence of certain epileptogenic lesions (eg, neurocysticercosis). As an example, emergency department-based studies include a larger number with acute symptomatic seizures that are more likely to have corresponding CT or MRI abnormalities [13]. Also, older patient populations are more likely to have structural brain lesions identified on MRI as a cause for epilepsy than are populations of primarily younger adults.

While brain MRI is routinely used in the clinical evaluation of epilepsy, there remains a substantial gap between the sensitivity and specificity of MRI in different care settings [10,17]. In one study, 123 patients referred for epilepsy surgery evaluation had had a "standard" MRI at an outside center (not attached to an epilepsy center) [17]. The original MRI interpretation was compared with a reinterpretation of the original MRI and with the results of a repeat MRI performed using a dedicated epilepsy protocol (see 'Epilepsy protocol for MRI' below). The sensitivity of MRI for focal lesions for each of these was 39, 50, and 91 percent, respectively. This was a selected population with refractory epilepsy that excluded patients with acute focal brain conditions, including cerebral abscess and rapidly expanding brain tumors. In most cases, the missed diagnosis was hippocampal sclerosis.

Epilepsy protocol for MRI — In 15 to 30 percent of patients with drug-resistant epilepsy, the brain MRI is negative; that is, no structural lesion can be identified [18-20]. To optimize the yield, MRI should be performed using an epilepsy protocol [6,21,22]. While epilepsy protocols vary depending on the institution and available technology, most recommendations agree that an epilepsy protocol for MRI should ideally include:

●Standard T1-weighted images.

●Three-dimensional (3D) volumetric T1-weighted images (1 mm isotropic voxels) with high definition of the gray-white junction (eg, magnetization-prepared rapid acquisition gradient echo [MPRAGE] and 3D fast spoiled gradient recalled echo acquisition at steady state [3D fast spoiled GRASS or 3D-SPGR]) for evaluation of brain anatomy, detection of malformations of cortical development, and application of postprocessing techniques such as 3D reconstructions and volumetric analyses.

●Axial and coronal T2-weighted (T2/short tau inversion recovery [STIR]) imaging for assessment of hippocampal architecture, basal temporal encephaloceles, and cystic tissue components of other lesions.

●Axial and coronal fluid-attenuated inversion recovery (FLAIR) sequences for assessing signal abnormalities and detection of hippocampal sclerosis, focal cortical dysplasia, tumors, inflammation, and scars.

●Axial T2 gradient echo or susceptibility-weighted sequences for identification of vascular and calcified lesions such as cavernomas and arteriovenous malformations, small hemosiderin deposits, and prior traumatic brain injury.

A widely accepted imaging protocol for epilepsy-specific imaging based on the above sequences was shown to identify 99.4 percent of 2740 epileptogenic lesions, providing a reasonable balance between diagnostic accuracy and clinical feasibility [23].

Imaging sequences should consist of contiguous, thin (<1.5 mm) slices that cover the entire brain. All the above sequences should be obtained in two orthogonal planes, with coronal images obtained obliquely and oriented perpendicular to the hippocampus in such a way that allows direct comparison between the left and right hemispheres. The oblique coronal orientation minimizes partial volume effects that otherwise commonly obscure hippocampal sclerosis and small lesions in the temporal lobe. The use of gadolinium-based contrast is not required for initial diagnostic MRI studies, but can be used to better characterize pathologies seen on noncontrast study or to improve sensitivity in initially negative studies [21,24].

MRI evidence of hippocampal atrophy is a strong predictor of excellent postoperative seizure control after anterior temporal lobectomy or laser interstitial thermal treatment [25]. For cases of hippocampal sclerosis where qualitative radiologic findings are not seen, however, quantitative volumetric hippocampal analysis may help to bridge the gap between sensitivity and specificity. Volumetry correlates well with histopathologically confirmed hippocampal cell loss [26]. Thus, reduced volume by quantitative analysis is an established surrogate marker for the presence and severity of hippocampal atrophy [27]. However, hippocampal volumetry has been difficult to incorporate in clinical practice because of the time demands and the technical skills required. Instead, automated software for generic quantitative morphometrics has substituted time-consuming manual techniques. Studies have shown that these automated techniques can detect hippocampal asymmetry and lateralize hippocampal atrophy accurately [28].

Advanced MRI techniques — Sensitivity also appears to be improved by more advanced MRI technologies. In particular, higher magnetic field strength (eg, 3-Tesla and 7-Tesla) and use of multichannel phased-array surface coils allow for a higher signal-to-noise ratio, improved image uniformity, and better spatial resolution [29]. One early study found that the use of phased-array surface coil MRI performed at 3-Tesla detected focal lesions in 65 percent of patients with a previously negative MRI [30]. Findings included cortical dysplasia, benign tumors, and hippocampal sclerosis. In one-third of patients with a previous abnormal MRI, the 3-Tesla MRI study further characterized lesional pathology and anatomic extent. Other reports also confirm the superiority of higher field-strength imaging study [31,32].

Diffusion tensor imaging (DTI) [33-35], magnetization transfer imaging [36], voxel-based analysis [37], and T2 mapping [38,39] are other technologies that show promise in the improved detection of both hippocampal sclerosis and malformations of cortical development. DTI provides information regarding the direction of the diffusion of water in each voxel, which can be used to estimate the orientation of white-matter tracts. This information can be used to trace major myelinated tracts (tractography) as part of a surgical evaluation; for example, it can be used for visualization of the optic radiation and for predicting visual field deficits after surgery [40,41]. (See "Surgical treatment of epilepsy in adults".)

Susceptibility-weighted imaging (SWI) is a three-dimensional MRI sequence with improved spatial resolution and enhanced magnetic susceptibility for compounds that distort the local magnetic field, such as blood products (ie, hemosiderin), deoxyhemoglobin in venous blood, iron content (ie, ferritin), and calcium content. SWI is more sensitive in detecting cavernous malformations compared with both T2-weighted FSE and T2*-weighted gradient-echo sequences [42]. This technique also appears to be helpful in identifying epileptogenic, postinfectious, and calcified lesions (eg, cryptococcus, tuberculosis, cysticercosis) [43].

High-resolution MRI is required for the diagnosis of many malformations of cortical development. On a standard MRI, findings suggestive of dysplasia include cortical thickening, blurring of the gray-white margin, increased signal on FLAIR, and subtle tapering bands of gray matter extending from the cortex toward the ventricles (image 3) [44]. Subtle lesions can be missed, and false-positive MRI readings can result from overinterpretation of normal variations in cortical thickness [24]. Though not yet widely in clinical use, 7-Tesla has even greater ability than 3-Tesla MRI to detect structural lesions using gradient echo and FLAIR images, and is especially promising for the detection of focal cortical dysplasias missed on conventional MRI [45-47].

Specificity of MRI findings for epilepsy — Not all MRI abnormalities are associated with epileptic seizures. Punctate foci of T2-hyperintense signal in the white matter, many cystic lesions (arachnoid cysts, choroidal fissure cysts), lacunar strokes, ventricular asymmetry, diffuse atrophy, and isolated venous anomalies (eg, developmental venous anomalies that are not associated with cavernous malformation) are not known to be epileptogenic, and should be considered incidental to a seizure diagnosis [24].

When a potentially epileptogenic structural abnormality is seen on brain MRI, it suggests an anatomic substrate for epilepsy, and provides support for a diagnosis of epilepsy. However, such findings should not be interpreted in isolation and must be correlated with the patient's seizure semiology and EEG findings. Several observations suggest that such findings might be incidental in a small proportion of patients.

●In patients with idiopathic generalized epilepsy, MRI abnormalities have been reported in up to 24 percent [48,49]. While most of these findings were not epileptogenic (eg, arachnoid cyst, diffuse cortical atrophy, etc), potentially epileptogenic lesions were seen in 3 to 4 percent. Similarly, one series reported abnormal MRI findings in 15 percent of 71 children with clinical and EEG features of benign childhood epilepsy with central midtemporal spikes [50]. Less than half of these findings were potentially epileptogenic. The benign course of these children's epilepsy syndromes suggests that these findings were truly incidental.

●In patients with psychogenic nonepileptic seizures, an abnormal brain MRI is reported in 10 to 38 percent [51-55]. Again, some, but not all of these abnormalities were epileptogenic, such as post-traumatic gliosis and hippocampal sclerosis [54].

●In a study comparing brain MRIs in 51 healthy controls and 99 patients with temporal lobe epilepsy (TLE), increased signal was noted in one or both hippocampi in 29 percent of controls (compared with 47.5 percent of patients) [56]. Hippocampal atrophy was a more specific finding, noted in only one control subject versus 19 percent of patients; no control had both atrophy and increased signal in the same hippocampus. (See "Focal epilepsy: Causes and clinical features", section on 'Hippocampal sclerosis'.)

●In normal, healthy adult volunteers, intracranial abnormalities were observed in 4.2 percent in one study of 1000 individuals [57]. Pathologies included those that were potentially epileptogenic (eg, tumors, remote trauma), as well as those that were not (arachnoid cyst, empty sella, lacunar infarction). This study may have underestimated the prevalence of relevant abnormalities, especially mesial temporal sclerosis, since they were not performed with a dedicated epilepsy protocol.

●Amygdala enlargement (AE) (image 8) is increasingly being reported on MRI in patients with TLE, with higher incidences reported in nonlesional TLE (12 to 63 percent) than in TLE with mesial temporal sclerosis (14 percent) [58-63]. This has led to the suggestion that it represents a distinct TLE subtype [64]. However, AE is observed at high rates in the amygdala contralateral to seizure onset [58], and epileptogenic activity appears to arise from the hippocampus, not the enlarged amygdala, at least in some cases [65]. Small sample sizes and variable methods for defining AE make it difficult to evaluate whether AE is specific to TLE [66]. Histopathologically, AE may be characterized by clustering hypertrophic neurons and vacuolation with slight gliosis or even no gliosis [65].

●Temporal encephaloceles are parenchymal protrusions through a bony defect in the middle cranial fossa (image 9) [67-69]. They are a relatively rare cause of medically refractory epilepsy. Detection of encephaloceles is facilitated by thin-slice 3D MRI sequences and skull base CT. As with other lesions, a causal relationship between temporal encephaloceles and the epilepsy syndrome is not always clear, and the surgical approach to symptomatic lesions varies. (See "Surgical treatment of epilepsy in adults", section on 'Temporal encephaloceles'.)

Another potential source of diagnostic confusion is that some patients may have acute MRI changes that are attributed to the effect of seizures rather than their cause. These are more common after prolonged seizures (eg, status epilepticus) or clusters of seizures, and are characterized by focal cortical swelling, increased T2-FLAIR signal intensity, restricted diffusion, and focal parenchymal and/or leptomeningeal contrast enhancement that resolve on subsequent imaging studies. These are discussed separately. (See "Magnetic resonance imaging changes related to acute seizure activity".)

FUNCTIONAL MRI — Functional magnetic resonance imaging (fMRI) can detect focal changes in blood flow and oxygenation levels that occur when an area of the brain is activated. A change in the level of neuronal activity is accompanied by a change in the ratio of concentration of oxy- to deoxyhemoglobin in the blood measured on MRI as the blood-oxygen-level-dependent (BOLD) effect. fMRI can be used to noninvasively map eloquent cortex serving motor, sensory, and language functions (image 10), and is most commonly used as part of surgical planning to predict and limit postoperative neurologic deficits, particularly language function, though it may also be able to help predict memory outcomes [70-78].

A 2017 guideline and meta-analysis from the American Academy of Neurology (AAN) assessed the value of fMRI for patients with epilepsy in determining lateralization and predicting postsurgical language and memory outcomes [79]. The authors concluded that language lateralization based on fMRI was concordant with the Wada test in mesial temporal lobe epilepsy (concordance rate 87 percent) and in extratemporal lobe epilepsy (concordance rate 81 percent), but those data were insufficient for temporal tumors or lateral temporal cases. The guideline concluded that fMRI is possibly effective in predicting postsurgical language deficits in patients undergoing temporal lobectomy.

fMRI may eventually replace the more invasive carotid amobarbital (Wada) test, particularly for language lateralization [80,81]. Early work suggested that fMRI may be able to visualize the functional anatomy of memory tasks and assist decision-making and planning of epilepsy surgery [74,82-89]. In another study, lateralization of memory using a picture recognition paradigm predicted postoperative verbal and visual memory outcome independent of the type of lesion, the side of the epileptic focus, or the type of preoperative memory profile [90]. However, a 2023 systematic review and meta-analysis found that task-based fMRI was more accurate at predicting postsurgical cognitive outcomes in patients with left-sided, rather than right-sided, temporal lobe epilepsy [91]. (See "Surgical treatment of epilepsy in adults".)

Interpretation of fMRI requires caution; it is an indirect measure of brain function. Discrepancies with the Wada test have been described [92-95]. Its sensitivity and specificity are imperfect (84 and 88 percent compared with the Wada test, in one meta-analysis), particularly in extratemporal epilepsy, and analyses have not been standardized [96]. Also, some patients cannot complete fMRI or have inconclusive fMRI results and still require preoperative Wada testing [95]. There are good data on its ability to lateralize language but not localize it adequately for surgical planning [97]. Thus, correct and clinically useful interpretation of fMRI for presurgical evaluation strongly depends on the expertise of the individual investigator and center.

Simultaneous recording of EEG and fMRI can visualize the BOLD response during interictal or ictal epileptic activity. This emerging technique capitalizing on the spatial resolution of fMRI and the temporal resolution of EEG could assist in identifying targets for surgical treatment [98-107]. However, the utility of fMRI for this purpose is not yet established.

PET — 18-F fluorodeoxyglucose positron emission tomography (FDG-PET) demonstrates the topographic distribution of glucose uptake in the brain and provides a picture of cerebral metabolism (image 11). FDG-PET is obtained during interictal period since cerebral uptake occurs 30 to 40 minutes after injection and thus, unless a seizure is long-lasting, an area of uptake may not be detected during the ictus. The goal of interictal FDG-PET is to identify focal areas of decreased metabolism (ie, relative hypometabolism) that are presumed to reflect focal functional disturbances of cerebral activity associated with epileptogenic tissue, otherwise known as the functional deficit zone. FDG-PET is generally performed as part of a presurgical evaluation.

In mesial temporal lobe epilepsy (MTLE), FDG-PET shows a widespread ipsilateral hypometabolism involving the mesial temporal structures, temporal pole, and lateral temporal cortex and often involving extratemporal areas including the insula, the frontal lobe, perisylvian regions, and the thalamus [108]. Given that the region of FDG-PET hypometabolism can extend beyond the epileptogenic lesion, FDG-PET should not be used to define margins for epilepsy surgery [109]. Furthermore, hypometabolism in the extratemporal regions and/or contralateral hemisphere in patients with mesial temporal lobe epilepsy is associated with a poorer response to epilepsy surgery [110].

Sensitivity for detecting relative temporal lobe hypometabolism with FDG-PET in MTLE ranges between 80 and 90 percent [111-118]. Visual assessment is less accurate than voxel-based morphometry, normalized to reference templates of healthy controls [119,120]. Much of the variability in sensitivity reflects the heterogeneity of the epilepsy more than it does the differences in quality or specifications of the PET camera [121]. Some reports suggest that the sensitivity of PET is increased when seizures are more frequent or when performed soon after a seizure has occurred [122].

Only a few patients in the above series included patients with MTLE without hippocampal sclerosis on MRI. However, these and other series have shown that FDG-PET can be helpful in lateralizing the epileptogenic temporal lobe in "MRI-negative" cases, with a yield that ranges from 45 to almost 90 percent [112-116,123-126].

Hypometabolic patterns on FDG-PET are predictive for surgical outcome in patients with MTLE [110,127]. Specifically, Engel class IA postsurgical outcome (ie, completely seizure-free after surgery) is associated with a focal anteromesial temporal hypometabolism, whereas suboptimal nonclass IA outcomes correlate with extratemporal metabolic changes [110].

There is less information available regarding the usefulness of FDG-PET in extratemporal epilepsy, but it appears somewhat less sensitive in these cases [13,111,113,128-131]. An exception may be patients with cortical dysplasia, in whom the reported sensitivity of PET varies between 60 and 92 percent [132]. Another limitation of PET is that the area of hypometabolism typically extends beyond the epileptogenic zone, making it less useful for precise neuroanatomic localization [108,111,130,133]. Similarly, coregistration of FDG-PET and MRI shows potential for improved sensitivity and specificity compared with either technology alone [134-136]. This is particularly useful in MRI-negative extratemporal lobe epilepsy (ETLE) for detecting focal cortical dysplasias and thus significantly improving the diagnosis and surgical outcome of these patients [135].

The use of other tracers (eg, [11C]flumazenil, [18F]FET, [18F]DPA-714, alpha methyl tryptophan, and serotonin agonists) remains largely experimental but holds promise for improving the sensitivity and specificity of PET for presurgical localization [113,118,122,130,137-146].

SPECT — Neuronal activity is strongly correlated with regional cerebral blood flow (CBF) neurovascular coupling. As such, an assessment of regional CBF during a seizure (ictus) may enable localization of the seizure focus. In single-photon emission computed tomography (SPECT), regional CBF can be imaged using perfusion radiotracers 99mTc-hexamethylpropyleneamine-oxime (HMPAO) or 99mTc-ethyl cysteinate dimer (ECD). Rapid initial uptake of these radiotracers in brain peaks within the first minutes of injection. SPECT obtained after 30 to 90 minutes for HMPAO and 30 to 60 minutes for ECD (range 10 minutes to six hours) provides a snapshot of CBF at the time of tracer administration.

Optimal evaluation of the epileptogenic focus with SPECT requires acquisition of two scans, one during the seizure (ictal) and one at baseline (interictal). The ictal scan requires the patient to be admitted to the hospital and monitored electrographically and clinically for seizure activity [147]. For ictal SPECT, the radiotracer should ideally be administered within 20 seconds of electrographic seizure onset; injection after 45 seconds is more likely to be falsely or non-localizing [122,148,149]. Increased regional CBF during the ictal phase (image 12) is followed by decreased CBF; the entire cerebral lobe and contralateral hemisphere may exhibit reduced perfusion after the seizure [150,151]. The sensitivity of SPECT is improved by comparison of ictal and interictal SPECT studies with quantitative subtraction techniques or by statistical comparison with a control database (statistical parametric mapping) [133,152,153]. Coregistration with MRI, a technique known as subtraction ictal SPECT scan coregistered with MRI (SISCOM), improves localization and can help predict surgical success [152,154-158].

SPECT can be useful in the identification of a possible epileptic focus, particularly when MRI is unremarkable, that can be further tested with intracranial EEG studies [159,160]. In some cases, re-evaluation of MRI prompted by abnormal SPECT has revealed a subtle structural lesion [161]. Ictal SPECT studies have a high yield in the evaluation of temporal lobe epilepsy (TLE) that exceeds 90 percent, but the sensitivity for extratemporal seizure foci is lower [111,117,162-164]. In one study of 53 patients with secondary generalized seizures, ictal SPECT identified an unambiguous seizure focus in only 50 percent; however, when a single unambiguous region of cerebral blood flow increase was seen, this was the correct localization in 80 percent [160].

However, SPECT studies have certain limitations in their utility to localize the seizure focus. Injection timing is critical; the tracer takes 30 seconds to reach the brain and the switch from ictal hyperperfusion to post-ictal hypoperfusion can take place in less than one or two minutes [122,133,165]. Propagation of seizure activity from the original focus of seizure onset to other brain regions can be even quicker [166]. As a result, areas of ictal hyperperfusion on SPECT may represent a false localization demonstrating areas of seizure propagation rather than seizure onset [160,162]. This may make SPECT localization a less reliable predictor of surgical outcome compared with other imaging studies, at least in some case series [122,123]. One study suggested that an injection time of less than 20 seconds after seizure onset is an important predictor of accurate localization; technically, this can be difficult to execute [167].

MEG AND MSI — Magnetoencephalography (MEG) is the recording of magnetic fields generated by intraneuronal electrical currents. Magnetic source imaging (MSI) is the coregistration of MEG source localization with anatomic imaging, MRI in most cases [168].

MEG is similar to EEG. However, while the electrical currents that are measured with EEG are attenuated in strength and spatially blurred by tissues between the brain and the scalp surface, the magnetic fields assessed by MEG are not significantly affected by intervening tissue layers [169]. As a result, MEG may allow for more clinically reliable localization of brain activity [170].

MEG and EEG can be viewed as complementary studies [171]. MEG is maximally sensitive to dipoles situated tangentially to the surface, whereas EEG is maximally sensitive to radial dipoles [172]. Single spikes with small amplitudes (eg, from mesial temporal sources) can be averaged to increase their signal-to-noise ratio. Averaged MEG spikes can thus be helpful in detecting discharges that are hidden in the background noise in simultaneous EEG recordings (waveform 1 and image 13). This phenomenon is reciprocal: in some cases, EEG spikes are more apparent than on MEG [171]. Some studies suggest that MEG has specific advantages for spike detection in extratemporal epilepsies, particularly those that lie superficially on the brain's surface [173-175].

MEG/MSI have been approved for presurgical localization of epilepsy and may be particularly useful for localization of spike sources in the following patients [133,174,176]:

●Those with no lesion visible on MRI [174,177-182].

●Those with cystic lesions (post-traumatic encephalomalacia with prior surgical resection) to determine the relationship between the epileptogenic focus and the lesion; MEG/MSI also appears to be useful in identifying the extent of the epileptogenic lesion in patients with focal cortical dysplasia, cavernous angiomas, and tumors [183-187].

●Those with MRI lesions of undetermined significance, including those with dual pathology or multifocal pathology such as tuberous sclerosis and others [188-190].

●Those who have had previous unsuccessful epilepsy surgery [183,191,192].

A large study in 455 epilepsy patients demonstrated an average sensitivity for MEG of 70 percent for specific epileptic activity [193]. Among patients who underwent surgical therapy, MSI provided a localization in 89 percent, supplying additional information in 35 percent and information that was crucial to medical decision-making in 10 percent. Similar sensitivity has been reported in other studies [173,178,194-196].

MEG is also approved for the localization of neuronal function (similar to evoked potentials and functional MRI [fMRI]) for language, sensorimotor, or visual cortex, and has been used to localize other cortical functions as well. MEG language mapping has been shown to agree with results of the Wada test in 75 to 95 percent of patients [171,172,197-199].

ESI — Electrical source imaging (ESI) is a model-based imaging technique that integrates spatial and temporal components of EEG to identify the generators of abnormal electrical activity associated with seizures [200].

In electrical source localization, by reconstructing the electric potentials recorded with scalp EEG, the location of the underlying currents can be estimated and merged with structural images of individual patients. It is typically based on the analysis of interictal epileptiform discharges (IEDs) but can also be calculated from ictal EEG discharges [201,202].

ESI enhances noninvasive localization accuracy of the epileptogenic zone [200,203]. High density-based ESI using a high number (128 to 256) of recording EEG electrodes has further improved localization accuracy. In one study, the diagnostic performance of ESI compared favorably with that of MRI and PET [204]. Importantly, concordance of the ESI result and the epileptogenic lesion delineated by MRI was associated with a probability of being seizure free (positive predictive value) of 92 percent. ESI may be particularly helpful in patients being considered for intracranial monitoring who have no cortical lesions on MRI or who have neocortical epilepsies, large cortical lesions, or EEG discharges with bilateral synchrony [205].

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5^th to 6^th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10^th to 12^th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Basics topics (see "Patient education: Epilepsy in adults (The Basics)" and "Patient education: Epilepsy in children (The Basics)")

SUMMARY AND RECOMMENDATIONS — Neuroimaging is central to the evaluation of patients with epilepsy, especially in the identification of structural brain lesions that can serve as epileptogenic foci, and that might be surgically resectable if the patient becomes refractory to medical treatment.

●Because of its relative insensitivity compared with MRI, CT has a limited role in the evaluation of patients with seizures. It is usually restricted to the emergency department setting in the evaluation of acute symptomatic seizures. (See 'Computed tomography' above.)

●Virtually all patients with new-onset epilepsy should have an MRI study to identify a potential structural cause of epilepsy including hippocampal sclerosis, brain tumor, dysplasia, vascular malformation, temporal encephaloceles, and others. (See 'Specific pathologies' above.)

●While less than half of patients with epilepsy will have a cause identified on MRI, the sensitivity can be substantially improved by utilizing an epilepsy protocol. (See 'Sensitivity' above.)

●Not all MRI findings are relevant; isolated findings of diffuse atrophy, punctate foci of T2 signal abnormalities in the white matter, and other nonspecific findings are not known to be epileptogenic. MRI findings should be correlated with the patient's seizure semiology and EEG findings; some potentially epileptogenic lesions may be incidental. (See 'Specificity of MRI findings for epilepsy' above.)

●Other neuroimaging modalities, such as positron emission tomography (PET), single-photon emission computed tomography (SPECT), and magnetic source imaging (MSI) are primarily used in the presurgical evaluation of patients with medically refractory epilepsy. These are employed to better define the area of functional defect and epileptogenicity, to identify MRI-occult lesions, to identify the more active lesion in patients with dual or multiple pathologies, and to map neurologic functions as part of surgical planning. (See 'PET' above and 'SPECT' above and 'MEG and MSI' above.)

ACKNOWLEDGMENT — The editorial staff at UpToDate acknowledges Lawrence J Hirsch, MD, who contributed to an earlier version of this topic review.