Neuroradiology/Head and Neck Imaging
Review
Epilepsy Imaging in Adults: Getting It Right
Audio Available | 
Share
OBJECTIVE. The purpose of this article is to describe an MRI protocol optimized for epilepsy evaluation, common causes of epilepsy visualized on MR images of patients evaluated for medically intractable partial epilepsy, and the basic concepts of advanced imaging techniques in the evaluation of epilepsy.
CONCLUSION. Epilepsy is one of the most common neurologic disorders in the United States. The long-term seizure-free success of epilepsy surgery is related to the ability to define and completely resect the epileptogenic zone. Detection of structural lesions at preoperative imaging requires not only a dedicated epilepsy protocol but also meticulous examination of the images by the interpreting radiologist with particular attention to subtle abnormalities that might otherwise go unreported.
Epilepsy is the fourth most common nontraumatic neurologic disorder in the United States, following only migraine, cerebrovascular disease, and Alzheimer disease in prevalence [1]. An estimated 2.2 million Americans have epilepsy, and the incidence is nearly 150,000 new cases annually. One in 26 people will receive a diagnosis of epilepsy during their lives [2]. Worldwide, 50 million people have epilepsy, and the World Health Organization estimates an associated morbidity of nearly 7.5 million disability-adjusted life years (health years lost) by 2015 [3].
The primary goal of epilepsy surgery centers is to render patients seizure free without complications. Thirty to forty percent of patients have epilepsy refractory to pharmacologic management, and many of these patients can be effectively treated with epilepsy surgery [4]. The success of epilepsy surgery is directly correlated with the ability to completely resect the epileptogenic zone, defined as the minimum amount of cortex that must be resected to provide seizure freedom. High-resolution MRI is necessary to anatomically define macroscopic epileptogenic lesions. The epileptogenic zone includes the seizure-onset zone—as defined by findings at invasive and noninvasive electroencephalography, ictal SPECT, or magnetic source imaging—and a surrounding potential seizure onset zone [5].
Epilepsy surgery is more effective in providing seizure freedom when a structural epileptogenic lesion is identified on preoperative images and completely resected at surgery. Selective resections are preferable, because resection of nonlesional functional tissue adversely affects postoperative cognitive and memory function [6]. Depending on the series, 20–30% of patients with temporal epilepsy and 20–40% of patients with extratemporal epilepsy have no clear lesion visible on MR images [7].
A long-term prospective study of epilepsy surgery patients and control patients not treated surgically [8] showed that most of the patients were seizure free at follow-up 5–10 years after resective surgery and that many of these patients were able to successfully discontinue antiepilepsy drugs. Long-term seizure freedom was better after temporal lobe resection than after extratemporal resection. Conversely, none of the nonoperatively treated patients had sustained seizure freedom or were able to discontinue antiepilepsy drugs.
| Optimizing Lesion Detection |   |
|---|
The success of epilepsy surgery is directly correlated with the ability to define and subsequently resect the epileptogenic zone. Seizure-free outcome is superior in patients who have focal circumscribed lesions present on presurgical MR images compared with patients in whom these lesions are not present [9]. A meta-analysis showed that the odds of becoming seizure free after surgery were 2.5 times higher in patients with defined lesions present on MR images or at histopathologic examination [7]. Improved lesion detection requires both a dedicated epilepsy protocol for MRI optimized for lesion detection and the interpreting radiologist's familiarity with common causes of epilepsy. The common causes of epilepsy discernible with imaging are hippocampal sclerosis (HS), congenital or developmental malformation, tumor, stroke, trauma, infection, vascular malformation, meningoencephalocele, hypoxicischemic encephalopathy, phakomatosis, and inborn error of metabolism.
To the extent that patient safety allows, images should be acquired with a 3-T MRI system. Examinations performed at 3 T have a better signal-to-noise ratio and better contrast resolution of the gray-white matter junction than do those performed at 1.5 T. Both lesion detection and characterization are improved with 3-T MRI [10] (Fig. 1). Even more important is the use of multichannel phased-array surface coils, which allow a higher signal-to-noise ratio, improved image uniformity, and better spatial and contrast resolution than does conventional quadrature head coil imaging [11, 12]. Such enhancements facilitate detection of cortical lesions, especially subtle cortical dysplasia. In a study of 40 consecutively registered patients with medically refractory focal epilepsy, 65% of patients (15 of 23) with normal findings at 1.5-T standard head coil brain MRI had new findings at 3-T eight-channel phased-array brain MRI. Most of the new findings were malformations of cortical development, most commonly focal cortical dysplasia (FCD) [13].
 View larger version (450K) | Fig. 1A —21-year-old woman with complex partial seizures. Comparison of 1.5-T and 3-T MRI. A, Axial T2-weighted MR images obtained with 1.5-T (A) and 3-T (B) systems show polymicrogyria in right temporal lobe (arrows, B) and nodular subependymal heterotopia (arrowhead, B) more clearly on 3-T image. |
 View larger version (436K) | Fig. 1B —21-year-old woman with complex partial seizures. Comparison of 1.5-T and 3-T MRI. B, Axial T2-weighted MR images obtained with 1.5-T (A) and 3-T (B) systems show polymicrogyria in right temporal lobe (arrows, B) and nodular subependymal heterotopia (arrowhead, B) more clearly on 3-T image. |
 View larger version (263K) | Fig. 1C —21-year-old woman with complex partial seizures. Comparison of 1.5-T and 3-T MRI. C, Sagittal T1-weighted images obtained with 1.5-T (C) and 3-T (D) systems show polymicrogyria along sylvian fissure and superior temporal sulcus (arrows, D) more clearly on 3-T images. |
 View larger version (258K) | Fig. 1D —21-year-old woman with complex partial seizures. Comparison of 1.5-T and 3-T MRI. D, Sagittal T1-weighted images obtained with 1.5-T (C) and 3-T (D) systems show polymicrogyria along sylvian fissure and superior temporal sulcus (arrows, D) more clearly on 3-T images. |
High-resolution coronal oblique T2-weighted MR images (2- to 3-mm slice thickness, no interspace gap) should be acquired along a plane perpendicular to the hippocampal long axis. Problematically, the anteroposterior FOV in this sequence is often limited to the hippocampal region. It is crucial that a high-resolution coronal T2-weighted sequence extend through the entire brain. Subtle cortical abnormalities may be present outside a limited hippocampal FOV, and an orthogonal plane is often helpful for identifying or confirming such anomalies. High-resolution coronal T2-weighted imaging through the anterior aspect of the middle cranial fossa is helpful for detecting temporal lobe encephalocele, a commonly overlooked and treatable cause of temporal lobe epilepsy (TLE).
Coronal T2-weighted FLAIR images (3-mm slice thickness, no interspace gap) are useful for confirming hippocampal T2 signal abnormalities and for detecting cortical and subcortical signal abnormalities related to FCD. Coronal FLAIR images, particularly in combination with a coronal 3D inversion recovery sequence, more clearly depict some subtle focal abnormalities, such as bottom-ofsulcus dysplasia. Cortical thickening, gray-white matter interface blurring, and associated signal abnormalities may be more apparent on coronal than on axial images, depending on sulcal orientation [14]. My colleagues and I routinely perform a thin-slice 3D T2-weighted FLAIR Cube sequence (1-mm isotropic voxels, sagittal acquisition), and the images can then be reformatted in any plane. In extratemporal epilepsies, it may be helpful to acquire images in planes tangential and perpendicular to the abnormal gyrus. In patients with partial seizures, the aura or clinical symptoms may indicate the region of the brain where the seizures are generated [15].
Compared with conventional MRI sequences, T2*-weighted gradient-recalled echo (GRE) or susceptibility weighted imaging (SWI) increases sensitivity for focal epileptogenic lesions by enhancing detection of calcifications, hemorrhage, and small occult vascular malformations [16]. Acquisition with a coronally oriented T2* GRE sequence improves detection of small lesions near the skull base, which may otherwise be obscured by susceptibility artifact.
Three-dimensional T1-weighted GRE isotropic sequences (magnetization prepared rapid acquisition GRE [MPRAGE] or spoiled gradient-recalled acquisition [SPGR]) with high spatial resolution (1-mm isotropic voxels) provide excellent contrast between gray and white matter and facilitate high-resolution assessment of cortical thickness. We routinely perform an MPRAGE sequence in the axial plane and reformat the data into the coronal and sagittal planes. The volumetric acquisition allows reformatting in additional planes to elucidate subtle malformations of cortical development. In addition, disturbances in cortical migration may become apparent, despite being inconspicuous on MR images obtained with conventional sequences (Fig. 2). Three-dimensional acquisition sequences can also be used to perform hippocampal segmentation and volume measurements [17].
 View larger version (376K) | Fig. 2A —21-year-old man with cortical migrational abnormality. A, Axial T2-weighted MR image shows no apparent abnormality. |
 View larger version (392K) | Fig. 2B —21-year-old man with cortical migrational abnormality. B, Axial T1-weighted magnetization-prepared rapid acquisition GRE image shows symmetric band (arrows) of abnormal heterotopic subcortical gray matter signal intensity. |
In most cases, evaluation of chronic seizures that have not changed in frequency or characteristics does not generally warrant use of IV contrast agents. Notable exceptions are patients with known or suspected enhancing tumors or neurocutaneous syndromes. New-onset seizures in an adult require contrast-enhanced imaging in addition to routine MRI sequences [18]. Indiscriminate use of IV contrast material in the evaluation of seizures in young children is not indicated, unless there is a known or suspected history of infection, neoplasm, or phakomatosis [19].
| Temporal Lobe Epilepsy: Beyond Mesial Temporal Sclerosis | |
|---|
The presence or absence of HS is one of the most consistently reported findings in MRI performed for evaluation of TLE. This is for good reason, because HS is by far the most common cause of TLE, present in 60–80% of surgical and autopsy specimens of patients with TLE [20]. Characteristic findings include T2-hyperintense gliosis in addition to atrophy and loss of normal internal morphologic macrostructure of the involved hippocampus (Fig. 3A). Preoperative MRI reliably depicts HS in more than 95% of patients [21]. In advanced cases, other portions of the ipsilateral limbic system, including the mammillary body and fornix, entorhinal cortex, and amygdala, may be atrophic [22].
 View larger version (362K) | Fig. 3A —Temporal lobe epilepsy, hippocampal sclerosis, and white matter abnormalities in anterior temporal lobe. A, 6-year-old girl with complex partial seizures and hippocampal sclerosis. Coronal T2-weighted FLAIR MR image shows increased T2 signal intensity in sclerotic hippocampus (arrow). |
 View larger version (434K) | Fig. 3B —Temporal lobe epilepsy, hippocampal sclerosis, and white matter abnormalities in anterior temporal lobe. B, 21-year-old woman with temporal lobe epilepsy and anterior temporal lobe epilepsy and anterior temporal lobe white matter abnormalities. Axial (B) and coronal (C) T2-weighted images show blurring at gray-white matter junction (arrows, B) and white matter volume loss and increased signal intensity (arrow, C) in anterior left temporal lobe. |
 View larger version (490K) | Fig. 3C —Temporal lobe epilepsy, hippocampal sclerosis, and white matter abnormalities in anterior temporal lobe. C, 21-year-old woman with temporal lobe epilepsy and anterior temporal lobe epilepsy and anterior temporal lobe white matter abnormalities. Axial (B) and coronal (C) T2-weighted images show blurring at gray-white matter junction (arrows, B) and white matter volume loss and increased signal intensity (arrow, C) in anterior left temporal lobe. |
Difficulty may arise in cases of bilateral HS. In such cases, T2 relaxometry is useful for visualizing concurrent involvement of the contralateral hippocampus [23]. HS may rarely exhibit atrophy in the absence of changes in T2 signal intensity [15].
Temporopolar changes are commonly present in TLE but not consistently reported. Such abnormalities include increased T2 signal intensity in the white matter, loss of gray-white matter distinction, and volume loss, particularly in the white matter core of the anterior temporal lobe (Figs. 3B and 3C). Temporal pole abnormalities are coexistent in most patients with HS but can also be an isolated finding in TLE. White matter T2 signal intensity abnormalities in the anterior temporal lobe are predictive of the side of seizure foci and are associated with a significantly greater extent of temporopolar white matter volume loss [24, 25].
Dual pathologic mechanisms—the coexistence of HS with extrahippocampal pathologic findings—are estimated to occur in approximately 15% of cases. They are most commonly related to congenital lesions or lesions that develop early in life, presumably reflecting increased vulnerability of the hippocampus in early childhood. The most common extrahippocampal abnormalities are cortical malformations, such as dysplasia and heterotopia [26]. The best postoperative seizure-free outcome occurs with resection of both pathologic entities [27].
Ipsilateral hippocampal inversion describes a morphologic variation of the hippo-campus that occurs when the normal process of inversion and infolding of the developing hippocampus is not completed. Consequently, the hippocampus assumes a round or pyramidal shape, the collateral sulcus is vertically oriented, and the collateral white matter is lateral to the hippocampus as opposed to below it [28] (Fig. 4). Ipsilateral hippocampal inversion is a common morphologic variation, not infrequently found in healthy persons. In patients with epilepsy, ipsilateral hippocampal inversion may reflect altered brain development affecting other parts of the brain but is not itself a cause of epilepsy [29].
 View larger version (381K) | Fig. 4 —9-year-old boy with ipsilateral hippocampal inversion. Coronal T2-weighted MR image shows pyramidal shape of left hippocampus related to abnormally configured and more vertically oriented collateral sulcus (arrowhead) compared with normal appearance of right collateral sulcus (asterisk). |
| Meningoencephaloceles | |
|---|
Middle cranial fossa encephaloceles are an underreported cause of TLE and are a more common cause of epilepsy than the rarity often ascribed to them in the literature would indicate. Failure to detect meningoencephaloceles may be attributed to limitations imposed by the imaging parameters used, lack of meticulous inspection of the anterior temporal lobe and skull base, or a combination of these factors. Given the typical location of meningoencephaloceles at the anterior aspect of the middle cranial fossa and that many of the small meningoencephaloceles are most conspicuous on coronal T2-weighted images, failure of high-resolution coronal T2-weighted MRI to include the entire brain may partially or completely exclude these lesions.
Meningoencephaloceles can be single or multiple and range in size from radiographically occult lesions, which are detectable only at surgery, to very large lesions. Middle cranial fossa encephaloceles commonly occur through bony defects in the greater wing of the sphenoid or lateral wall of the sphenoid sinus (Fig. 5A). Thin-slice CT with multiplanar reformations is helpful for confirming associated bony abnormalities, which may range from pitting defects in the greater sphenoid wing, resembling changes secondary to arachnoid granulations, to regions of frank osseous dehiscence (Fig. 5B). Meningoencephaloceles may occur elsewhere in the brain, including more posteri-orly in the temporal lobe and along the floor of the anterior cranial fossa (Fig. 5C).
 View larger version (355K) | Fig. 5A —Meningoencephaloceles. A, 33-year-old woman with tonic-clonic seizures and multiple small temporal lobe. Coronal oblique T2-weighted MR image through anterior aspect of middle cranial fossa shows multiple small meningoencephaloceles (arrowheads). |
 View larger version (332K) | Fig. 5B —Meningoencephaloceles. B, 33-year-old woman with tonic-clonic seizures and multiple small temporal lobe. Coronal CT image through skull base shows cluster of bone defects in left greater sphenoid wing related to temporal encephaloceles (arrow). |
 View larger version (307K) | Fig. 5C —Meningoencephaloceles. C, 58-year-old woman with frontal encephalocele. Sagittal T1-weighted image shows frontoethmoidal encephalocele (arrow). Failure to inspect entire skull base for defects can lead to delay in diagnosis. |
| Focal Epileptogenic Lesions: Tumors Versus Dysplasia | |
|---|
Tumors are present in 25–35% of pathologic specimens obtained at operations for chronic epilepsy. With rare exceptions, epilepsy-associated tumors are almost always low grade (World Health Organization grade I or II). Brain tumors associated with medically intractable epilepsy differ in frequency from the common tumors found in general neuro-surgical series. Gangliogliomas, dysembryo-plastic neuroepithelial tumors (DNETs), and pleomorphic xanthoastrocytomas (PXAs) are more commonly found in patients with medically intractable epilepsy. Other common epilepsy-associated tumors are low-grade gliomas (oligodendrogliomas and astrocytomas) and low-grade mixed glial tumors and mixed glial-glioneuronal tumors [30].
Ganglioglioma is the most common brain tumor in patients with medically intractable epilepsy across multiple series [21, 30, 31] (Fig. 6A). Approximately one half of gangliogliomas are predominantly solid masses, and approximately one half are primarily cystic. Most cystic lesions have associated solid components. The solid components are typically T2 hyperintense. Approximately 40% of gangliogliomas have calcifications, and less than one half are enhancing on MR images [32]. DNETs are commonly multicystic cortex-based masses, and internal areas of cystic fluid signal intensity are suppressed on T2 FLAIR images (Fig. 6B). DNETs may also comprise a single cyst or contain solid elements, which are enhancing in approximately one half of cases. On CT images, DNETs are hypoattenuating masses with calcifications uncommonly present [33]. PXAs are rare cortex-based, T2-hyperintense tumors classically described as a cyst with a mural nodule abutting the meninges, although they may be solid in at least one half of cases. PXAs typically are enhancing, often with associated leptomeningeal enhancement (Figs. 6C and 6D). Calcification is rare [34]. Surrounding vasogenic edema is uncommon in these low-grade lesions.
 View larger version (346K) | Fig. 6A —Epilepsy-associated tumors. A, 13-year-old girl with ganglioglioma. Axial contrast-enhanced T1-weighted MR image shows large cystic lesion in left parietooccipital region with enhancing mural nodule (arrow). |
 View larger version (421K) | Fig. 6B —Epilepsy-associated tumors. B, 12-year-old boy with dysembryoplastic neuroepithelial tumor. Axial T2-weighted FLAIR image shows bubbly mass (arrow) in anterior right temporal lobe with high-signal-intensity rim and internal septations. No surrounding edema is present. |
 View larger version (411K) | Fig. 6C —Epilepsy-associated tumors. C, 21-year-old woman with pleomorphic xanthoastrocytoma. Axial T2-weighted (C) and sagittal contrast-enhanced T1-weighted (D) images show T2-hyperintense mass (arrow, C) in mesial left temporal lobe with characteristic meningocerebral enhancement. Dural enhancement extends along tentorium (arrowheads, D). Minimal surrounding edema is present. |
 View larger version (309K) | Fig. 6D —Epilepsy-associated tumors. D, 21-year-old woman with pleomorphic xanthoastrocytoma. Axial T2-weighted (C) and sagittal contrast-enhanced T1-weighted (D) images show T2-hyperintense mass (arrow, C) in mesial left temporal lobe with characteristic meningocerebral enhancement. Dural enhancement extends along tentorium (arrowheads, D). Minimal surrounding edema is present. |
Given that many low-grade tumors are not enhancing, the distinction between low-grade tumor and cortical dysplasia is not always clear at routine imaging (Fig. 7A). FCD may also be coexistent with glioneuronal tumors, further obscuring the distinction. The presence of mass effect suggests neoplasm. The previously described epilepsy-associated tumors most commonly occur in the temporal lobes, whereas type II FCD (Taylor FCD), one of the most commonly surgical treated FCDs, is most often extratemporal in location, commonly found in the frontal lobes. Unfortunately, lesion location is not a reliable distinguishing characteristic. Type II FCD is characterized by focal cortical thickening, blurring of the gray-white matter junction, and subcortical T2 hyperintensity that appears as hypointensity on T1-weighted images [35] (Fig. 7B). Funnel-shaped tapering of the subcortical signal abnormality toward the ventricle (transmantle sign) is more commonly associated with FCD type IIb (balloon cell subtype) [36]. Mild T2 hyperintensity is sometimes apparent in the cortex [35]. Cortical thickening should be confirmed in two planes by use of high-resolution technique. The presence of gray-white matter blurring and acquisition of slices at an oblique orientation with respect to the gyrus may falsely accentuate cortical thickness assessment [36]. Some FCDs are commonly overlooked because of small size or location at the bottom of a sulcus [14]. Meticulous attention to cortical anatomy is necessary to detect subtle gyral pattern abnormalities.
 View larger version (421K) | Fig. 7A —Examples of Taylor IIb focal cortical dysplasia. A, 25-year-old woman with epilepsy. Axial T2-weighted MR image shows T2-hyperintense cortical thickening in left frontal lobe with adjacent white matter cystic change (arrow). Imaging findings may reflect cortical dysplasia or low-grade neoplasm. Biopsy results revealing gliosis were deemed to reflect nonrepresentative sampling. |
 View larger version (289K) | Fig. 7B —Examples of Taylor IIb focal cortical dysplasia. B, 11-year-old boy with type IIb focal cortical dysplasia. Coronal T2-weighted FLAIR image (B) shows focal cortical thickening and increased T2 signal intensity in left frontal lobe (arrow), correlating with focal region of cortical hypometabolism on interictal FDG PET scan (C). |
 View larger version (227K) | Fig. 7C —Examples of Taylor IIb focal cortical dysplasia. C, 11-year-old boy with type IIb focal cortical dysplasia. Coronal T2-weighted FLAIR image (B) shows focal cortical thickening and increased T2 signal intensity in left frontal lobe (arrow), correlating with focal region of cortical hypometabolism on interictal FDG PET scan (C). |
| Transient Seizure-Related MRI Signal Changes | |
|---|
Reversible MRI signal changes in the brain after prolonged seizure activity or status epilepticus can mimic other pathologic conditions. Such conditions may manifest unilateral or bilateral T2 hyperintensity and restricted diffusion of the cerebral cortex or subcortical white matter in addition to gyral swelling and evidence of increased relative cerebral blood flow on perfusion images [37]. Affected cortical regions are not confined to vascular territories. The hippocampus is commonly affected, and signal changes may also occur in the ipsilateral pulvinar region of the thalamus and rarely in the contralateral cerebellum [37, 38] (Fig. 8). Gyral enhancement may occur after contrast administration [39]. At follow-up imaging, DWI and T2 signal changes and cortical swelling normalize over time [38]. Some patients experience permanent brain injury, including cortical volume loss, laminar necrosis, and HS [39–41].
 View larger version (270K) | Fig. 8 —57-year-old man with status epilepticus. Diffusion-weighted image shows restricted diffusion in pulvinar region of left thalamus (arrow) and left temporal and occipital cortex (arrowheads) in patient with complex partial status epilepticus. Restricted diffusion after prolonged seizure activity is not restricted to single vascular territory resolves on follow-up imaging. |
| Vascular Malformations | |
|---|
Among adults with first-time seizures in the absence of hemorrhage or focal neurologic deficit, persons with cavernous malformations have a higher 5-year risk of development of epilepsy than those with arteriovenous malformations. Cavernous malformations are more likely to be multiple in adults with seizures [42]. Cavernous malformations have a characteristic appearance with a reticulated mixed-signal-intensity core consisting of hemorrhage of variable ages and a complete surrounding hemosiderin rim. They are particularly conspicuous on images obtained with GRE or SWI sequences. Certain anatomic characteristics of arteriovenous malformations have been statistically associated with a clinical presentation of epilepsy, including cortical location of the nidus, arterial supply by the middle cerebral or external carotid arteries, cortical location of the feeders, absence of aneurysm, and a temporal or parietal superficial location [43].
| Posttraumatic Epilepsy, Cerebrovascular Disease, and Infection | |
|---|
Traumatic brain injuries, cerebrovascular disease, and CNS infections are some of the most common predisposing causes of epilepsy. Cerebrovascular disease is the most common cause of acquired epilepsy in adults in Western populations [44]. Imaging of these conditions is fairly straightforward, and findings should be apparent on routine-protocol brain MR images. The presence of blood products in the cortex is known to be highly epileptogenic, attributed to iron deposition [45]. Traumatic brain injuries are frequently multifocal and bilateral, with contusions commonly occurring in the temporal lobes. SWI is more sensitive than imaging with conventional GRE sequences in depicting small hemorrhagic lesions [46].
Survivors of viral encephalitis and bacterial meningitis have a statistically significant increased risk of epilepsy, and this risk increases with early seizures [44]. Worldwide, neurocysticercosis is the most common parasitic infection of the human CNS. Although seizures are reported to be the most common clinical presentation, the clinical link between neurocysticercosis and epilepsy is inconsistent [47, 48]. The severity of epilepsy does not correlate with the number of cysticercosis lesions, and the epileptogenic zone can be spatially remote from the location of the infection. Parenchymal neurocysticercosis progresses through four radio-graphically distinct stages: vesicular stage with a CSF-isointense cyst and internal scolex, colloidal vesicular stage with proteinaceous cyst fluid and pericystic edema and enhancement, granular nodular stage at which the cyst begins to retract, and calcified nodular stage with an involuted, calcified granuloma [47].
| Advanced Imaging | |
|---|
Functional imaging of patients with epilepsy can be performed with radiolabeled tracers, as in 18F-FDG PET and 99mTc-exa met a zime or 99mTc-bicisate SPECT. The physiologic basis for these techniques is the general tendency for cortical glucose metabolism and blood flow to be increased in the epileptogenic focus during a seizure and decreased in the postictal and interictal periods [49]. Interictal FDG PET depicts the cortical area of interictal dysfunction as a focal region of glucose hypometabolism (Fig. 7C). In TLE, FDG PET is more sensitive than MRI for localization of the focus of a temporal lobe seizure. PET may lateralize the affected temporal lobe in almost one half of TLE cases with noncontributory electroencephalographic findings [50]. In patients with HS, the extent or severity of hypo metabolism on FDG PET images does not correlate with the extent of neuronal loss detected at histopathologic examination or MRI.
Some limitations of FDG PET should be noted. Because of the length of the uptake phase, approximately 40 minutes, findings obtained from FDG injection immediately before a seizure show a complex metabolic pattern related to both ictal and postictal changes [51]. The functional deficit zone is usually larger than the epileptogenic zone [50]. An epileptogenic focus may uncommonly be hypermetabolic at FDG PET owing to seizures soon after FDG injection, subclinical seizures, or physiologic cellular changes [49]. Ictal SPECT and interictal PET have similar accuracies for epileptogenic focus localization [51]. In evaluation of diffuse cerebral or lobar abnormalities, FDG PET can localize the site of greatest metabolic disturbance [49]. Retrospective review of the MR images after nuclear imaging may reveal a subtle cortical abnormality overlooked at the initial MRI interpretation (Fig. 9).
 View larger version (311K) | Fig. 9A —36-year-old woman with type IIa focal cortical dysplasia. A, FDG PET scan shows focal hypermetabolism in posterior left insular cortex (arrow). Patient had seizure soon after FDG injection, and hypermetabolism reflects ictal changes. |
 View larger version (356K) | Fig. 9B —36-year-old woman with type IIa focal cortical dysplasia. B, Axial T2-weighted MR image shows subtle focal cortical thickening in region (arrow) shown in A. PET abnormality enhances detection of subtle cortical lesions on conventional MR images. |
Although not a part of routine epilepsy imaging evaluations, proton MR spectroscopy may play an adjunctive role in the presurgical evaluation of certain patients with epilepsy. Multiple technical strategies can be used. Two- or three-dimensional magnetic spectroscopic imaging (multivoxel) point-resolved spectroscopy (PRESS) can be performed with a TE of 135 milliseconds [52]. A unilateral decrease in N-acetyl aspartate (NAA) to creatine (Cr) ratio, NAA/choline (Cho) ratio, or NAA/Cho + Cr ratio shows high concordance with electroencephalographic findings in lateralizing the epileptogenic focus in patients with TLE, and MR spectroscopy may therefore be helpful for confirming electroencephalographic abnormalities in patients with abnormal findings on conventional MR images [53, 54]. MR spectroscopic interpretation often relies on comparing the suspected diseased area with a contralateral ROI [53]. Incorrect lateralization of TLE is more likely to occur in cases of bilateral abnormalities [55].
Magnetic source imaging, which combines magnetoencephalography with structural MRI, can guide depth electrode placement in patients with a potential epileptogenic lesion that is not apparent on the MR images or with surface electroencephalographic findings that are discordant [56]. Magnetoencephalography is used to measure magnetic fields generated by small intracellular neuronal electrical currents. Dipole source modeling of waveforms localizes the epileptic focus [57] (Fig. 10).
 View larger version (243K) | Fig. 10A —27-year-old man with intractable partial epilepsy and nonlesional MRI findings. A, Magnetic source imaging (MSI) with magnetoencephalographic data overlayedon coronal (A), sagittal (B), and axial (C) T1-weighted MRI and surface-rendered model (D). MSI shows dipole clusters localized to posterior aspect of superior frontal gyrus, which correlates with electroencephalographic findings. In patients with no lesion identified on MRI, MSI can guide depth electrode placement. |
 View larger version (286K) | Fig. 10B —27-year-old man with intractable partial epilepsy and nonlesional MRI findings. B, Magnetic source imaging (MSI) with magnetoencephalographic data overlayedon coronal (A), sagittal (B), and axial (C) T1-weighted MRI and surface-rendered model (D). MSI shows dipole clusters localized to posterior aspect of superior frontal gyrus, which correlates with electroencephalographic findings. In patients with no lesion identified on MRI, MSI can guide depth electrode placement. |
 View larger version (352K) | Fig. 10C —27-year-old man with intractable partial epilepsy and nonlesional MRI findings. C, Magnetic source imaging (MSI) with magnetoencephalographic data overlayedon coronal (A), sagittal (B), and axial (C) T1-weighted MRI and surface-rendered model (D). MSI shows dipole clusters localized to posterior aspect of superior frontal gyrus, which correlates with electroencephalographic findings. In patients with no lesion identified on MRI, MSI can guide depth electrode placement. |
 View larger version (282K) | Fig. 10D —27-year-old man with intractable partial epilepsy and nonlesional MRI findings. D, Magnetic source imaging (MSI) with magnetoencephalographic data overlayedon coronal (A), sagittal (B), and axial (C) T1-weighted MRI and surface-rendered model (D). MSI shows dipole clusters localized to posterior aspect of superior frontal gyrus, which correlates with electroencephalographic findings. In patients with no lesion identified on MRI, MSI can guide depth electrode placement. |
In FCD, electroencephalographic abnormalities and magnetoencephalographic dipole clusters are often disseminated over a larger area than the extent of the MRI-delineated abnormality. Resection of the entire epileptogenic zone, which may extend beyond the margins of the MRI-visible abnormality, is necessary for the greatest chance of seizure freedom. Diffusion-tensor imaging has been found to show microstructural changes in the subcortical white matter adjacent to the MRI-visible abnormality and in the white matter tracts projecting to and from the FCD. Reduced anisotropy and increased diffusivity have been found in the white matter subjacent to the dysplastic cortex both within and adjacent to magnetoencephalographic dipole clusters, probably at least partly related to abnormal myelin. Diffusion-tensor imaging may play a role in defining the true extent of FCD [58].
| Conclusion | |
|---|
Seizure-free outcome after epilepsy surgery is worse when a structural lesion is not identified on the preoperative MR images [7]. An effective imaging strategy requires a dedicated epilepsy protocol, preferably performed with a 3-T system, and meticulous attention on the part of the interpreting radiologist to subtle abnormalities. Advanced imaging may provide localizing information when findings at conventional MRI are normal. Imaging detection of an epileptogenic lesion allows resection of the epileptogenic zone, improving the odds of long-term seizure freedom.
| References | |
|---|
