- Email: [email protected]
Neuroimaging in Partial Epilepsy: Structural Magnetic Resonance Imaging
Review Article
Neuroimaging in Partial Epilepsy: Structural Magnetic Resonance Imaging Gregory D. Cascino
Magnetic resonance imaging (MRI) is a noninvasive technique that has been shown to be the structural neuroimaging procedure of choice in evaluating patients with partial or localization-related epilepsy. The diagnostic yield of MRI has been confirmed in patients with partial epilepsy related to mesial temporal sclerosis (MTS) or foreign-tissue lesions. Magnetic resonance imaging may be used preoperatively to identify patients with intractable partial epilepsy who have surgically remediable epileptic syndromes. Preoperative MRI studies are predictive of long-term seizure outcome in patients receiving surgical treatment. Analysis of hippocampal formation size has also been shown to correlate with the neurocognitive outcome following temporal lobe surgery. A recent development involving subtraction ictal single photon emission computed tomography (SPECT) coregistered with structural MRI (SISCOM) has important clinical applications. SISCOM studies are more sensitive and specific than visual side-by-side interpretation of interictal and ictal SPECT scans. Also, SISCOM images have been shown to have prognostic importance in patients undergoing surgical treatment for epilepsy. Key Words: Magnetic resonance imaging—Lesional epilepsy—Hippocampal formation atrophy—SISCOM. © 1998 by Elsevier Science Inc. All rights reserved.
The last 25 years has witnessed the development of neuroimaging procedures to delineate the pathologic substrate underlying the epileptic brain tissue that corresponds to the site of seizure onset (1,2).
Received December 15, 1997; accepted December 15, 1997. From the Department of Neurology, Mayo Clinic, Rochester, MN, U.S.A. Address correspondence and reprint requests to Dr. G. D. Cascino, Division of Epilepsy, Department of Neurology, Mayo Clinic, 200 First Street, S.W., Rochester, MN 55905, U.S.A. J. Epilepsy 1998;11:121–129 © 1998 by Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
Early imaging techniques in epilepsy included conventional skull X-rays, cerebral arteriography, and pneumoencephalography (1,2). These modalities were of limited utility because of a low diagnostic yield or potential morbidity, or both (3). The introduction of X-ray computed tomography (CT) began a new era in the diagnostic evaluation of patients with epilepsy. Computed tomography proved useful in demonstrating “macroscopic” intracranial lesions associated with epilepsy (1–3). The potential limitations of CT in epilepsy included poor resolution of “microscopic” pathology, inability to perform multiplanar imaging, radiation exposure, dif-
0896-6974/98/$19.00 PII S0896-6974(98)00010-3
G. D. CASCINO
ficulty distinguishing gray and white matter structures, and bony artifact (4).
Introduction to MRI Magnetic resonance imaging (MRI) has been demonstrated to be the most sensitive and specific structural neuroimaging procedure in patients with partial or localization-related epilepsy (4). Importantly, MRI is a noninvasive technique that has no known biologic toxicity and does not involve ionizing radiation. Magnetic resonance imaging has been shown to be a reliable indicator of the pathology underlying the epileptogenic zone (i.e., site of seizure onset and initial seizure propagation) in patients with symptomatic partial epilepsy (4,5). The presence of an MRI-identified structural abnormality may suggest the localization of the site of seizure onset (4 – 8). The high diagnostic yield of MRI to delineate foreign-tissue lesions (e.g., tumor or vascular malformation) has been confirmed (Figs. 1–3) (4,6). There is also a broad consensus that MRI will identify mesial temporal sclerosis (MTS), the most common pathology encountered in patients with intractable partial epilepsy (Figs. 4 and 5) (5,7). Findings of MRI have been used to select favorable candidates for epilepsy surgery, tailor the operative resection, and confirm the extent of corticectomy postoperatively (4 –12).
Symptomatic Partial Epilepsy The three major categories of symptomatic partial epilepsy are as follows: (1) medial temporal lobe epilepsy, (2) lesional epilepsy, and (3) nonlesional neocortical epilepsy (1). Approximately 80% of patients with partial epilepsy have temporal lobe seizures (1,5–7). The epileptogenic zone in the temporal lobe involves the amygdalohippocampal complex in nearly 90% of patients (15). The pathologic hallmark of medial temporal lobe epilepsy is MTS (5,7). Surgically excised pathology in approximately 65% of patients with intractable partial epilepsy is MTS associated with focal neuronal loss with or without gliosis (5,7). A structural lesion underlying the epileptogenic zone is identified in approximately 30% of patients undergoing epilepsy surgery (1). Lesional pathology includes primary brain tumors, vascular malformations, and disorders of cortical development (8 –10,13). 122 J EPILEPSY, VOL. 11, NO. 3, 1998
Figure 1. (a) MRI head (T1-weighted, axial) reveals a neocortical left temporal lobe lesion (see arrow); (b) MRI head (T2-weighted, axial) in this patient demonstrates a signal intensity alteration related to the mass lesion. A ganglioglioma was excised.
Structural MRI Methodology in Epilepsy The MRI seizure protocol at the Mayo Clinic Rochester includes: (1) sagittal T1-weighted imaging with minimum echo time (TE) and 500 msec repetition time (TR) required for whole-head coverage with 5-mm thick contiguous sections; (2) whole-head coronal three-dimensional volumetric spoiled gradient-echo (SPGR) acquisition is per-
NEUROIMAGING IN EPILEPSY
Figure 2. MRI head (FLAIR, coronal) shows a left medial occipital tumor. T1-weighted and T2-weighted studies were unremarkable. A low-grade oligodendroglioma was excised.
formed with minimum full TE and TR, 192 views, one repetition, 1.5 mm section thickness with 124 partitions, 22 cm field of view, and 45 flip angle; and (3) coronal spin-echo (SE) imaging is performed with TE of 30 and 80 msec, TR greater than 2000 msec, 20 cm field of view, 4 mm section thickness and 2 mm intersection gap, and 192 views with one repetition (5,8,11,12). An oblique-coronal fluid attenuated inversion recovery (FLAIR) sequence is also obtained (14). The FLAIR sequence allows the pathologic signal change to be differentiated from the physiologic signal alteration related to cerebrospinal fluid (CSF) (Fig. 4). An enhanced study will be performed if a space-occupying lesion is detected in the unenhanced study.
MRI and Surgical Pathology
Figure 3. (a) MRI head (T2-weighted, oblique-coronal) shows a left temporal lobe cavernous hemangioma; (b) MRI head (T1-weighted, sagittal) in this patient.
Mesial Temporal Sclerosis Magnetic resonance findings in patients with MTS include hippocampal formation atrophy (HFA) and an increased mesial temporal signal intensity (Figs. 4 and 5) (5,7,11,14 –16). Inspection of MRI will allow detection of 80%–90% of the cases of MTS (5,7,11,14 –16). The HFA atrophy is most obvious using the T1-weighted image in the obliquecoronal plane (5,7). The signal intensity alteration can be identified using T2-weighted imaging or the
FLAIR sequence in the oblique-coronal plane. The coronal or oblique-coronal planes are useful for MRI studies in patients with MTS because of the capability to compare the two hippocampi for any side to side asymmetry (5,7,11,14,15). Potential limitations of visual inspection of the MRI in patients with suspected MTS includes the following: head rotation, symmetrical bilateral HFA, subtle unilateral HFA, or signal intensity alteration. Most imJ EPILEPSY, VOL. 11, NO. 3, 1998 123
G. D. CASCINO
Figure 4. MRI head (T1-weighted, oblique-coronal) indicates left hippocampal formation atrophy in a patient with mesial temporal sclerosis (see arrow).
portantly, visual inspection is a subjective determination that is strongly dependent on the inspector’s expertise for appropriate interpretation. Three-dimensional SPGR images are helpful because they are reformated into true anatomic coronal plane. Magnetic resonance imaging-based hippocampal formation volumetric studies have been developed to objectively determine the degree of hippocampal volume loss in patients with MTS (5,11,16). Absolute hippocampal volume measurements are performed using a standardized protocol with the results being compared to age-matched normal controls to assign abnormal values (5,11,16). A unilateral reduction in hippocampal volume has been shown to be a reliable indicator of the temporal lobe
of seizure origin in patients with medically refractory partial epilepsy. Jackson described T2 relaxometry to objectively determine the medial temporal lobe signal intensity (14). Quantitative MRI studies have limited clinical application because of the high diagnostic yield of visual inspection. The most important use of hippocampal formation volumetry and T2 relaxometry is for research studies. Hippocampal formation atrophy correlates with an early age of seizure onset, a history of a febrile seizure of childhood, and the diagnosis of medial temporal lobe epilepsy (17). A history of a neurologic illness in childhood (e.g., febrile seizure, head trauma, meningitis) appears to be an important risk factor for the development of MTS (17). The duration of epilepsy and age at the time of surgery have not correlated with volumetric results in most studies (17). The identification of MTS in the surgically excised temporal lobe has been a favorable prognostic indicator of seizure control following epilepsy surgery (11). Jack et al. have shown that nearly 90% of patients with unilateral hippocampal atrophy have been rendered seizure-free (11). Cascino et al. demonstrated that approximately 80% of patients with unilateral HFA coexistent with the epileptic brain tissue were rendered seizure-free after surgical treatment compared to only 50% with normal MRI studies (p , 0.0001) (18). Magnetic resonance imaging is now recognized as being predictive of neurocognitive outcome in patients undergoing an anterior temporal lobectomy (19). Patients with normal left hippocampal volumes are at greater risk for experiencing a significant decline in cognitive performance following a left medial temporal lobe resection than those with left HFA (19).
Primary Brain Neoplasms
Figure 5. MRI head (FLAIR, coronal) reveals a signal intensity alteration in the left hippocampal formation. Mesial temporal sclerosis was pathologically verified (see arrow). 124 J EPILEPSY, VOL. 11, NO. 3, 1998
The epileptogenic potential of primary brain neoplasms has been recognized since the initial work of J. Hughlings Jackson over one century ago (20). Dr. Jackson documented the relationship between the site of tumor and the ictal onset zone (20). The ictal behavior was shown to be related to the anatomic localization of the epileptic brain tissue and not the histopathology of the tumor (20). Low-grade, slowly growing tumors are most commonly associated with a chronic seizure disorder (20). The histopathology includes the following tumors: oligodendroglioma, fibrillary astrocytoma, pilocytic astrocytoma, mixed glioma, ganglioglioma, and a dysembroblastic neuroepithelial tumor (DNET) (21). The DNET probably is more closely related to
NEUROIMAGING IN EPILEPSY
Figure 6. MRI head (FLAIR, oblique-coronal) shows extensive changes related to tuberous sclerosis.
a disorder of cortical development than a primary brain neoplasm. Imaging features common to all these tumors include a typically small size, localization at or near a cortical surface, sharply defined borders, little or no surrounding edema and, with the exception of the pilocytic astrocytomas, little or no contrast enhancement (Figs. 1 and 2) (4,13). Most patients with primary central nervous system (CNS) tumors and chronic epilepsy do not present with progressive neurologic deficits, “changing”
Figure 7. MRI head (oblique-coronal) shows the “double cortex” or band heterotopia.
Figure 8. MRI head (T1-weighted, oblique-coronal) in a patient with the bilateral perisylvian syndrome.
neurologic examination, or evidence for increased intracranial pressure (21).
Vascular Malformations There are four types of congenital cerebral vascular malformations: arteriovenous malformations (AVM), cavernous hemangiomas, venous angiomas, and telangiectasias (13). The epileptogenic potential of these pathologic lesions is quite different.
Figure 9. MRI head (T1-weighted, oblique-coronal) reveals a large extratemporal neuronal migrational abnormality. J EPILEPSY, VOL. 11, NO. 3, 1998 125
The common malformations resected for partial epilepsy are cavernous hemangiomas and AVMs. Seizures may in fact be the only clinical manifestation with these lesions. Venous angiomas and telangiectasias are often incidental findings in patients with seizure disorders. Magnetic resonance imagings is essential for the recognition and diagnosis of cavernous hemangiomas and occult AVMs, which may be associated with a flow signal on MRI. Cavernous hemangiomas characteristically have a target appearance on T2-weighted images with a region of increased T2 signal intensity surrounded by an area of decreased signal produced by remote (often occult) hemorrhage with methemoglobin deposition (Fig. 3) (13). Resection typically leads to complete or significant improvement in seizure control (22–24).
Disorders of Cortical Development
Figure 10. MRI head showing three dimensional surface anatomy in a patient with focal cortical dysplasia (see arrow). The proposed craniotomy site is outlined in yellow (top). The actual operative photograph is superimposed on the MRI (bottom). F and H represent the sites of face and hand motor activity, respectively, as determined by electrical stimulation mapping.
Disorders of cortical development are an important etiology for symptomatic partial epilepsy (6,7,25,26). The use of MRI has allowed recognition of these lesions and demonstrated the frequency and importance of cortical developmental malformations in patients with intractable partial epilepsy (Figs. 6 –10). A variety of developmental abnormalities have been recognized that are commonly associated with medically refractory seizures and neurocognitive decline (6,7,25–27). Cerebral devel-
Figure 11. SPECT and SISCOM studies in a patient with right posterior frontal lobe epilepsy. Structural MRI was normal. SISCOM shows the region of altered cerebral perfusion in the medial aspect of the right cerebral hemisphere (see arrow). 126 J EPILEPSY, VOL. 11, NO. 3, 1998
NEUROIMAGING IN EPILEPSY
Figure 12. MRI and SISCOM studies in a patient with left temporal lobe epilepsy.
opmental malformations could previously only be diagnosed by postmortem examination. For the diagnosis and proper classification of these pathologic lesions, MRI is essential. Central nervous system insults can produce a cerebral developmental
malformation between the fifth and 10th weeks of gestation when the telencephalon is developed until the 25th week when cell migration is completed (6,7,25,26). A classification of the developmental abnormali-
Figure 13. Substraction SPECT and MRI three-dimensional surface map of the right cerebral hemisphere. Structural MRI did not reveal an abnormality. Selected intracranial electrodes positions are depicted. J EPILEPSY, VOL. 11, NO. 3, 1998 127
G. D. CASCINO
ties has been introduced by Kuzniecky based on the structural neuroimaging findings (27). The disorders of cortical development have been classified by localization of the pathologic lesions: generalized or diffuse disorders, unilateral disorders, focal disorders, and diffuse or focal disorders (27) (Figs. 6 –10). The generalized disorders include lissencephaly, “double cortex” or band heterotopia, subependymal heterotopia, and megalencephaly (Fig. 7). Hemimegalencephaly is a unilateral cortical developmental malformation. Focal disorders include the following: focal cortical dysplasia, polymicrogyria, schizencephaly, and focal subcortical heterotopia (Figs. 9 and 10). Focal cortical dysplasias are the most often considered for surgical resection. Thin section three-dimensional volumetric MRI is extremely useful in evaluating these anomalies (Fig. 10). It is difficult to resolve volume-averaged normal cortical infolding from true areas of abnormalities if the spatial resolution of the images is coarser than 1.5 mm. Reformatting of 1.5 mm three-dimensional SPGR MRI sequences is also helpful in this regard. Tuberous sclerosis is considered a diffuse or focal developmental abnormality (Fig. 6).
Subtraction SPECT and MRI Single photon emission computed tomography (SPECT) has been used in patients with intractable partial epilepsy to identify a cerebral perfusion alteration that may indicate the localization of the epileptic brain tissue (28 –30). There are conflicting findings regarding the diagnostic yield of interictal studies in patients undergoing a presurgical evaluation with medically refractory partial seizures. There is a consensus that ictal SPECT studies are more sensitive and specific than interictal examinations for lateralizing the site of seizure onset (28). Computer-aided subtraction of the interictal from the ictal SPECT images followed by coregistration to the MRI (SISCOM) is a recent development that has been shown to be diagnostically superior to routine visual inspection of the interictal and ictal scans (Figs. 11–13) (28 –30). O’Brien et al. at Mayo Clinic showed that visual side-by-side interpretation of the interictal and ictal SPECT scans proved less sensitive and specific than SISCOM in 51 patients with intractable partial epilepsy (Fig. 11) (28,30). Blinded interpretation of the SPECT scans were performed by two investigators. The SISCOM images revealed a localized alteration in cerebral perfusion in 45 of 51 patients (88.2%) compared to 128 J EPILEPSY, VOL. 11, NO. 3, 1998
20 of 51 patients (39.2%) using visual side-by-side interpretation (p , 0.0001) (28). The SISCOM studies were also shown to be more specific as determined by concordance with the long-term EEG monitoring (28). Finally, patients with SISCOM images that revealed a localized abnormality were more likely to experience an excellent outcome following epilepsy surgery than individuals with nonconcordant or nonlocalizing findings (p , 0.001) (28,29).
References 1. Cascino GD. Commentary: How has neuroimaging improved patient care. Epilepsia 1994;35(suppl 6):S103–7. 2. McRae DL. Radiology in epilepsy. In: Vinken PJ, Bruyn GW, eds. Handbook of clinical neurology. New York: Elsevier, 1974:553– 63. 3. Gastaut H. Conclusions: Computed transverse axial tomography in epilepsy. Epilepsia 1976;17:337– 8. 4. Bergen D, Bleck T, Ramsey R, et al. Magnetic resonance imaging as a sensitive and specific predictor of neoplasms removed for intractable epilepsy. Epilepsia 1989;30:310 –21. 5. Cascino GD, Jack CR Jr., Parisi JE, et al. Magnetic resonance imaging-based volume studies in temporal lobe epilepsy: Pathological correlation. Ann Neurol 1991;30:31– 6. 6. Barkovitch A, Chuang S, Norman D. MR of neuronal migration anomalies. AJNR 1987;8:1009 –17. 7. Berkovic S, Andermann F, Olivier A, et al. Hippocampal sclerosis in temporal lobe epilepsy demonstrated by MRI. Ann Neurol 1991;29:175– 82. 8. Cascino GD, Jack CR Jr., Parisi JE, et al. MRI in the presurgical evaluation of patients with frontal lobe epilepsy and children with temporal lobe epilepsy: Pathological correlation and prognostic importance. Epilepsy Res 1992;11:51–9. 9. Cascino GD, Jack CR Jr., Parisi JE, et al. Operative strategy in patients with MRI-identified dual pathology and temporal lobe epilepsy. Epilepsy Res 1993;14:175– 82. 10. Cascino GD, Jack CR Jr., Sharbrough FW, Kelly PJ, Marsh WR. MRI assessments of hippocampal pathology in patients with extratemporal lesional epilepsy. Neurology 1993;43: 2380 –2. 11. Jack CR Jr., Sharbrough FW, Cascino GD, Hirschorn KA, O’Brien PC, Marsh WR. Magnetic resonance imaging-based hippocampal volumetry: Correlation with outcome after temporal lobectomy. Ann Neurol 1992;31:138 – 46. 12. Jack CR Jr., Sharbrough FW, Twomey CK, et al. Temporal lobe seizures: Lateralization with MR volume measurements of hippocampal formation. Radiology 1990;176:205–9. 13. Friedland RJ, Bronen RA. Magnetic resonance imaging of neoplastic, vascular, and indeterminant substrates. In: Cascino GD, Jack CR Jr., eds. Neuroimaging in epilepsy: Principles and practice. Boston: Butterworth-Heinemann, 1996: 29 –50. 14. Jackson GD, Berkovic SF, Tress BM, et al. Hippocampal sclerosis can be reliably detected by magnetic resonance imaging. Neurology 1990;40:1869 –75. 15. Jackson GD. Visual analysis in mesial temporal sclerosis. In: Cascino GD, Jack CR Jr., eds. Neuroimaging in epilepsy: Principles and practice. Boston: Butterworth-Heinemann, 1996:73–110.
NEUROIMAGING IN EPILEPSY 16. Lenz T, McCarthy G, Bronen R, et al. Quantitative MRI of the hippocampus in temporal lobe epilepsy. Ann Neurol 1992; 31:629 –37. 17. Trenerry MR, Jack CR Jr., Cascino GD, et al. MRI hippocampal volumes: Association with onset and duration of epilepsy, and febrile convulsions in temporal lobectomy patients. Epilepsy Res 1993;15:247–52. 18. Cascino GD, Trenerry MR, So EL, et al. Routine EEG and temporal lobe epilepsy: Relation to long-term EEG monitoring, quantitative MRI, and operative outcome. Epilepsia 1996;37:651– 6. 19. Trenerry MR, Jack CR Jr, Cascino GD, et al. MRI hippocampal volumes and memory before and after temporal lobectomy. Neurology 1993;33:1800 –5. 20. Cascino GD, Epilepsy and brain tumors: Implications for treatment. Epilepsia 1990;31:S37–S44. 21. Britton J, Cascino GD, Sharbrough FW, Kelly PJ. Low-grade glial neoplasms and intractable partial epilepsy: Efficacy of surgical treatment. Epilepsia 1994;35:1130 –5. 22. Dodick D, Cascino GD, Meyer FB. Vascular malformations and intractable epilepsy: Outcome after surgical treatment. Mayo Clin Proc 1994;69:741–5. 23. Cascino GD, Kelly PJ, Sharbrough FW, Hirschorn KA. Longterm follow-up of stereotactic lesionectomy in partial epilepsy: Predictive factors and electroencephalographic results. Epilepsia 1992;33:659 – 44.
24. Fish DR, Andermann F, Olivier A, et al. Complex partial seizures and posterior temporal or extratemporal lesions: surgical strategies. Neurology 1991;41:1718 – 84. 25. Kuzniecky R, Murro A, King D, et al. Magnetic resonance imaging in childhood intractable partial epilepsies: Pathological correlation. Neurology 1993;43:681–7. 26. Palmini A, Andermann A, Olivier A, et al. Focal neuronal migrational disorders and intractable partial epilepsy: A study of 30 patients. Ann Neurol 1991;30:741–9. 27. Kuzniecky RI. Magnetic resonance imaging in cerebral developmental malformations and epilepsy. In: Cascino GD, Jack CR Jr, eds. Neuroimaging in epilepsy: Principles and practice. Boston: Butterworth-Heinemann, 1996:51– 63. 28. O’Brien TJ, So EL, Mullan BP, et al. Subtraction ictal SPECT co-registered to MRI improves clinical usefulness of SPECT in localizing the surgical seizure focus. Neurology 1998;50:445– 454. 29. O’Brien TJ, So EL, Mullan BP, et al. Extent of resection of the ictal subtraction SPECT focus is an important determinant of epilepsy surgery outcome. Epilepsia 1996; 37(Suppl 5):S182. 30. So EL, O’Brien TJ, Mullan BP, et al. Seizure localization in intractable partial epilepsy by subtraction SPECT co-registered to MRI: A study of 47 consecutive patients. Epilepsia 1996;37(Suppl 5):S124.
J EPILEPSY, VOL. 11, NO. 3, 1998 129
Neuroimaging in Partial Epilepsy: Structural Magnetic Resonance Imaging Gregory D. Cascino
Magnetic resonance imaging (MRI) is a noninvasive technique that has been shown to be the structural neuroimaging procedure of choice in evaluating patients with partial or localization-related epilepsy. The diagnostic yield of MRI has been confirmed in patients with partial epilepsy related to mesial temporal sclerosis (MTS) or foreign-tissue lesions. Magnetic resonance imaging may be used preoperatively to identify patients with intractable partial epilepsy who have surgically remediable epileptic syndromes. Preoperative MRI studies are predictive of long-term seizure outcome in patients receiving surgical treatment. Analysis of hippocampal formation size has also been shown to correlate with the neurocognitive outcome following temporal lobe surgery. A recent development involving subtraction ictal single photon emission computed tomography (SPECT) coregistered with structural MRI (SISCOM) has important clinical applications. SISCOM studies are more sensitive and specific than visual side-by-side interpretation of interictal and ictal SPECT scans. Also, SISCOM images have been shown to have prognostic importance in patients undergoing surgical treatment for epilepsy. Key Words: Magnetic resonance imaging—Lesional epilepsy—Hippocampal formation atrophy—SISCOM. © 1998 by Elsevier Science Inc. All rights reserved.
The last 25 years has witnessed the development of neuroimaging procedures to delineate the pathologic substrate underlying the epileptic brain tissue that corresponds to the site of seizure onset (1,2).
Received December 15, 1997; accepted December 15, 1997. From the Department of Neurology, Mayo Clinic, Rochester, MN, U.S.A. Address correspondence and reprint requests to Dr. G. D. Cascino, Division of Epilepsy, Department of Neurology, Mayo Clinic, 200 First Street, S.W., Rochester, MN 55905, U.S.A. J. Epilepsy 1998;11:121–129 © 1998 by Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
Early imaging techniques in epilepsy included conventional skull X-rays, cerebral arteriography, and pneumoencephalography (1,2). These modalities were of limited utility because of a low diagnostic yield or potential morbidity, or both (3). The introduction of X-ray computed tomography (CT) began a new era in the diagnostic evaluation of patients with epilepsy. Computed tomography proved useful in demonstrating “macroscopic” intracranial lesions associated with epilepsy (1–3). The potential limitations of CT in epilepsy included poor resolution of “microscopic” pathology, inability to perform multiplanar imaging, radiation exposure, dif-
0896-6974/98/$19.00 PII S0896-6974(98)00010-3
G. D. CASCINO
ficulty distinguishing gray and white matter structures, and bony artifact (4).
Introduction to MRI Magnetic resonance imaging (MRI) has been demonstrated to be the most sensitive and specific structural neuroimaging procedure in patients with partial or localization-related epilepsy (4). Importantly, MRI is a noninvasive technique that has no known biologic toxicity and does not involve ionizing radiation. Magnetic resonance imaging has been shown to be a reliable indicator of the pathology underlying the epileptogenic zone (i.e., site of seizure onset and initial seizure propagation) in patients with symptomatic partial epilepsy (4,5). The presence of an MRI-identified structural abnormality may suggest the localization of the site of seizure onset (4 – 8). The high diagnostic yield of MRI to delineate foreign-tissue lesions (e.g., tumor or vascular malformation) has been confirmed (Figs. 1–3) (4,6). There is also a broad consensus that MRI will identify mesial temporal sclerosis (MTS), the most common pathology encountered in patients with intractable partial epilepsy (Figs. 4 and 5) (5,7). Findings of MRI have been used to select favorable candidates for epilepsy surgery, tailor the operative resection, and confirm the extent of corticectomy postoperatively (4 –12).
Symptomatic Partial Epilepsy The three major categories of symptomatic partial epilepsy are as follows: (1) medial temporal lobe epilepsy, (2) lesional epilepsy, and (3) nonlesional neocortical epilepsy (1). Approximately 80% of patients with partial epilepsy have temporal lobe seizures (1,5–7). The epileptogenic zone in the temporal lobe involves the amygdalohippocampal complex in nearly 90% of patients (15). The pathologic hallmark of medial temporal lobe epilepsy is MTS (5,7). Surgically excised pathology in approximately 65% of patients with intractable partial epilepsy is MTS associated with focal neuronal loss with or without gliosis (5,7). A structural lesion underlying the epileptogenic zone is identified in approximately 30% of patients undergoing epilepsy surgery (1). Lesional pathology includes primary brain tumors, vascular malformations, and disorders of cortical development (8 –10,13). 122 J EPILEPSY, VOL. 11, NO. 3, 1998
Figure 1. (a) MRI head (T1-weighted, axial) reveals a neocortical left temporal lobe lesion (see arrow); (b) MRI head (T2-weighted, axial) in this patient demonstrates a signal intensity alteration related to the mass lesion. A ganglioglioma was excised.
Structural MRI Methodology in Epilepsy The MRI seizure protocol at the Mayo Clinic Rochester includes: (1) sagittal T1-weighted imaging with minimum echo time (TE) and 500 msec repetition time (TR) required for whole-head coverage with 5-mm thick contiguous sections; (2) whole-head coronal three-dimensional volumetric spoiled gradient-echo (SPGR) acquisition is per-
NEUROIMAGING IN EPILEPSY
Figure 2. MRI head (FLAIR, coronal) shows a left medial occipital tumor. T1-weighted and T2-weighted studies were unremarkable. A low-grade oligodendroglioma was excised.
formed with minimum full TE and TR, 192 views, one repetition, 1.5 mm section thickness with 124 partitions, 22 cm field of view, and 45 flip angle; and (3) coronal spin-echo (SE) imaging is performed with TE of 30 and 80 msec, TR greater than 2000 msec, 20 cm field of view, 4 mm section thickness and 2 mm intersection gap, and 192 views with one repetition (5,8,11,12). An oblique-coronal fluid attenuated inversion recovery (FLAIR) sequence is also obtained (14). The FLAIR sequence allows the pathologic signal change to be differentiated from the physiologic signal alteration related to cerebrospinal fluid (CSF) (Fig. 4). An enhanced study will be performed if a space-occupying lesion is detected in the unenhanced study.
MRI and Surgical Pathology
Figure 3. (a) MRI head (T2-weighted, oblique-coronal) shows a left temporal lobe cavernous hemangioma; (b) MRI head (T1-weighted, sagittal) in this patient.
Mesial Temporal Sclerosis Magnetic resonance findings in patients with MTS include hippocampal formation atrophy (HFA) and an increased mesial temporal signal intensity (Figs. 4 and 5) (5,7,11,14 –16). Inspection of MRI will allow detection of 80%–90% of the cases of MTS (5,7,11,14 –16). The HFA atrophy is most obvious using the T1-weighted image in the obliquecoronal plane (5,7). The signal intensity alteration can be identified using T2-weighted imaging or the
FLAIR sequence in the oblique-coronal plane. The coronal or oblique-coronal planes are useful for MRI studies in patients with MTS because of the capability to compare the two hippocampi for any side to side asymmetry (5,7,11,14,15). Potential limitations of visual inspection of the MRI in patients with suspected MTS includes the following: head rotation, symmetrical bilateral HFA, subtle unilateral HFA, or signal intensity alteration. Most imJ EPILEPSY, VOL. 11, NO. 3, 1998 123
G. D. CASCINO
Figure 4. MRI head (T1-weighted, oblique-coronal) indicates left hippocampal formation atrophy in a patient with mesial temporal sclerosis (see arrow).
portantly, visual inspection is a subjective determination that is strongly dependent on the inspector’s expertise for appropriate interpretation. Three-dimensional SPGR images are helpful because they are reformated into true anatomic coronal plane. Magnetic resonance imaging-based hippocampal formation volumetric studies have been developed to objectively determine the degree of hippocampal volume loss in patients with MTS (5,11,16). Absolute hippocampal volume measurements are performed using a standardized protocol with the results being compared to age-matched normal controls to assign abnormal values (5,11,16). A unilateral reduction in hippocampal volume has been shown to be a reliable indicator of the temporal lobe
of seizure origin in patients with medically refractory partial epilepsy. Jackson described T2 relaxometry to objectively determine the medial temporal lobe signal intensity (14). Quantitative MRI studies have limited clinical application because of the high diagnostic yield of visual inspection. The most important use of hippocampal formation volumetry and T2 relaxometry is for research studies. Hippocampal formation atrophy correlates with an early age of seizure onset, a history of a febrile seizure of childhood, and the diagnosis of medial temporal lobe epilepsy (17). A history of a neurologic illness in childhood (e.g., febrile seizure, head trauma, meningitis) appears to be an important risk factor for the development of MTS (17). The duration of epilepsy and age at the time of surgery have not correlated with volumetric results in most studies (17). The identification of MTS in the surgically excised temporal lobe has been a favorable prognostic indicator of seizure control following epilepsy surgery (11). Jack et al. have shown that nearly 90% of patients with unilateral hippocampal atrophy have been rendered seizure-free (11). Cascino et al. demonstrated that approximately 80% of patients with unilateral HFA coexistent with the epileptic brain tissue were rendered seizure-free after surgical treatment compared to only 50% with normal MRI studies (p , 0.0001) (18). Magnetic resonance imaging is now recognized as being predictive of neurocognitive outcome in patients undergoing an anterior temporal lobectomy (19). Patients with normal left hippocampal volumes are at greater risk for experiencing a significant decline in cognitive performance following a left medial temporal lobe resection than those with left HFA (19).
Primary Brain Neoplasms
Figure 5. MRI head (FLAIR, coronal) reveals a signal intensity alteration in the left hippocampal formation. Mesial temporal sclerosis was pathologically verified (see arrow). 124 J EPILEPSY, VOL. 11, NO. 3, 1998
The epileptogenic potential of primary brain neoplasms has been recognized since the initial work of J. Hughlings Jackson over one century ago (20). Dr. Jackson documented the relationship between the site of tumor and the ictal onset zone (20). The ictal behavior was shown to be related to the anatomic localization of the epileptic brain tissue and not the histopathology of the tumor (20). Low-grade, slowly growing tumors are most commonly associated with a chronic seizure disorder (20). The histopathology includes the following tumors: oligodendroglioma, fibrillary astrocytoma, pilocytic astrocytoma, mixed glioma, ganglioglioma, and a dysembroblastic neuroepithelial tumor (DNET) (21). The DNET probably is more closely related to
NEUROIMAGING IN EPILEPSY
Figure 6. MRI head (FLAIR, oblique-coronal) shows extensive changes related to tuberous sclerosis.
a disorder of cortical development than a primary brain neoplasm. Imaging features common to all these tumors include a typically small size, localization at or near a cortical surface, sharply defined borders, little or no surrounding edema and, with the exception of the pilocytic astrocytomas, little or no contrast enhancement (Figs. 1 and 2) (4,13). Most patients with primary central nervous system (CNS) tumors and chronic epilepsy do not present with progressive neurologic deficits, “changing”
Figure 7. MRI head (oblique-coronal) shows the “double cortex” or band heterotopia.
Figure 8. MRI head (T1-weighted, oblique-coronal) in a patient with the bilateral perisylvian syndrome.
neurologic examination, or evidence for increased intracranial pressure (21).
Vascular Malformations There are four types of congenital cerebral vascular malformations: arteriovenous malformations (AVM), cavernous hemangiomas, venous angiomas, and telangiectasias (13). The epileptogenic potential of these pathologic lesions is quite different.
Figure 9. MRI head (T1-weighted, oblique-coronal) reveals a large extratemporal neuronal migrational abnormality. J EPILEPSY, VOL. 11, NO. 3, 1998 125
The common malformations resected for partial epilepsy are cavernous hemangiomas and AVMs. Seizures may in fact be the only clinical manifestation with these lesions. Venous angiomas and telangiectasias are often incidental findings in patients with seizure disorders. Magnetic resonance imagings is essential for the recognition and diagnosis of cavernous hemangiomas and occult AVMs, which may be associated with a flow signal on MRI. Cavernous hemangiomas characteristically have a target appearance on T2-weighted images with a region of increased T2 signal intensity surrounded by an area of decreased signal produced by remote (often occult) hemorrhage with methemoglobin deposition (Fig. 3) (13). Resection typically leads to complete or significant improvement in seizure control (22–24).
Disorders of Cortical Development
Figure 10. MRI head showing three dimensional surface anatomy in a patient with focal cortical dysplasia (see arrow). The proposed craniotomy site is outlined in yellow (top). The actual operative photograph is superimposed on the MRI (bottom). F and H represent the sites of face and hand motor activity, respectively, as determined by electrical stimulation mapping.
Disorders of cortical development are an important etiology for symptomatic partial epilepsy (6,7,25,26). The use of MRI has allowed recognition of these lesions and demonstrated the frequency and importance of cortical developmental malformations in patients with intractable partial epilepsy (Figs. 6 –10). A variety of developmental abnormalities have been recognized that are commonly associated with medically refractory seizures and neurocognitive decline (6,7,25–27). Cerebral devel-
Figure 11. SPECT and SISCOM studies in a patient with right posterior frontal lobe epilepsy. Structural MRI was normal. SISCOM shows the region of altered cerebral perfusion in the medial aspect of the right cerebral hemisphere (see arrow). 126 J EPILEPSY, VOL. 11, NO. 3, 1998
NEUROIMAGING IN EPILEPSY
Figure 12. MRI and SISCOM studies in a patient with left temporal lobe epilepsy.
opmental malformations could previously only be diagnosed by postmortem examination. For the diagnosis and proper classification of these pathologic lesions, MRI is essential. Central nervous system insults can produce a cerebral developmental
malformation between the fifth and 10th weeks of gestation when the telencephalon is developed until the 25th week when cell migration is completed (6,7,25,26). A classification of the developmental abnormali-
Figure 13. Substraction SPECT and MRI three-dimensional surface map of the right cerebral hemisphere. Structural MRI did not reveal an abnormality. Selected intracranial electrodes positions are depicted. J EPILEPSY, VOL. 11, NO. 3, 1998 127
G. D. CASCINO
ties has been introduced by Kuzniecky based on the structural neuroimaging findings (27). The disorders of cortical development have been classified by localization of the pathologic lesions: generalized or diffuse disorders, unilateral disorders, focal disorders, and diffuse or focal disorders (27) (Figs. 6 –10). The generalized disorders include lissencephaly, “double cortex” or band heterotopia, subependymal heterotopia, and megalencephaly (Fig. 7). Hemimegalencephaly is a unilateral cortical developmental malformation. Focal disorders include the following: focal cortical dysplasia, polymicrogyria, schizencephaly, and focal subcortical heterotopia (Figs. 9 and 10). Focal cortical dysplasias are the most often considered for surgical resection. Thin section three-dimensional volumetric MRI is extremely useful in evaluating these anomalies (Fig. 10). It is difficult to resolve volume-averaged normal cortical infolding from true areas of abnormalities if the spatial resolution of the images is coarser than 1.5 mm. Reformatting of 1.5 mm three-dimensional SPGR MRI sequences is also helpful in this regard. Tuberous sclerosis is considered a diffuse or focal developmental abnormality (Fig. 6).
Subtraction SPECT and MRI Single photon emission computed tomography (SPECT) has been used in patients with intractable partial epilepsy to identify a cerebral perfusion alteration that may indicate the localization of the epileptic brain tissue (28 –30). There are conflicting findings regarding the diagnostic yield of interictal studies in patients undergoing a presurgical evaluation with medically refractory partial seizures. There is a consensus that ictal SPECT studies are more sensitive and specific than interictal examinations for lateralizing the site of seizure onset (28). Computer-aided subtraction of the interictal from the ictal SPECT images followed by coregistration to the MRI (SISCOM) is a recent development that has been shown to be diagnostically superior to routine visual inspection of the interictal and ictal scans (Figs. 11–13) (28 –30). O’Brien et al. at Mayo Clinic showed that visual side-by-side interpretation of the interictal and ictal SPECT scans proved less sensitive and specific than SISCOM in 51 patients with intractable partial epilepsy (Fig. 11) (28,30). Blinded interpretation of the SPECT scans were performed by two investigators. The SISCOM images revealed a localized alteration in cerebral perfusion in 45 of 51 patients (88.2%) compared to 128 J EPILEPSY, VOL. 11, NO. 3, 1998
20 of 51 patients (39.2%) using visual side-by-side interpretation (p , 0.0001) (28). The SISCOM studies were also shown to be more specific as determined by concordance with the long-term EEG monitoring (28). Finally, patients with SISCOM images that revealed a localized abnormality were more likely to experience an excellent outcome following epilepsy surgery than individuals with nonconcordant or nonlocalizing findings (p , 0.001) (28,29).
References 1. Cascino GD. Commentary: How has neuroimaging improved patient care. Epilepsia 1994;35(suppl 6):S103–7. 2. McRae DL. Radiology in epilepsy. In: Vinken PJ, Bruyn GW, eds. Handbook of clinical neurology. New York: Elsevier, 1974:553– 63. 3. Gastaut H. Conclusions: Computed transverse axial tomography in epilepsy. Epilepsia 1976;17:337– 8. 4. Bergen D, Bleck T, Ramsey R, et al. Magnetic resonance imaging as a sensitive and specific predictor of neoplasms removed for intractable epilepsy. Epilepsia 1989;30:310 –21. 5. Cascino GD, Jack CR Jr., Parisi JE, et al. Magnetic resonance imaging-based volume studies in temporal lobe epilepsy: Pathological correlation. Ann Neurol 1991;30:31– 6. 6. Barkovitch A, Chuang S, Norman D. MR of neuronal migration anomalies. AJNR 1987;8:1009 –17. 7. Berkovic S, Andermann F, Olivier A, et al. Hippocampal sclerosis in temporal lobe epilepsy demonstrated by MRI. Ann Neurol 1991;29:175– 82. 8. Cascino GD, Jack CR Jr., Parisi JE, et al. MRI in the presurgical evaluation of patients with frontal lobe epilepsy and children with temporal lobe epilepsy: Pathological correlation and prognostic importance. Epilepsy Res 1992;11:51–9. 9. Cascino GD, Jack CR Jr., Parisi JE, et al. Operative strategy in patients with MRI-identified dual pathology and temporal lobe epilepsy. Epilepsy Res 1993;14:175– 82. 10. Cascino GD, Jack CR Jr., Sharbrough FW, Kelly PJ, Marsh WR. MRI assessments of hippocampal pathology in patients with extratemporal lesional epilepsy. Neurology 1993;43: 2380 –2. 11. Jack CR Jr., Sharbrough FW, Cascino GD, Hirschorn KA, O’Brien PC, Marsh WR. Magnetic resonance imaging-based hippocampal volumetry: Correlation with outcome after temporal lobectomy. Ann Neurol 1992;31:138 – 46. 12. Jack CR Jr., Sharbrough FW, Twomey CK, et al. Temporal lobe seizures: Lateralization with MR volume measurements of hippocampal formation. Radiology 1990;176:205–9. 13. Friedland RJ, Bronen RA. Magnetic resonance imaging of neoplastic, vascular, and indeterminant substrates. In: Cascino GD, Jack CR Jr., eds. Neuroimaging in epilepsy: Principles and practice. Boston: Butterworth-Heinemann, 1996: 29 –50. 14. Jackson GD, Berkovic SF, Tress BM, et al. Hippocampal sclerosis can be reliably detected by magnetic resonance imaging. Neurology 1990;40:1869 –75. 15. Jackson GD. Visual analysis in mesial temporal sclerosis. In: Cascino GD, Jack CR Jr., eds. Neuroimaging in epilepsy: Principles and practice. Boston: Butterworth-Heinemann, 1996:73–110.
NEUROIMAGING IN EPILEPSY 16. Lenz T, McCarthy G, Bronen R, et al. Quantitative MRI of the hippocampus in temporal lobe epilepsy. Ann Neurol 1992; 31:629 –37. 17. Trenerry MR, Jack CR Jr., Cascino GD, et al. MRI hippocampal volumes: Association with onset and duration of epilepsy, and febrile convulsions in temporal lobectomy patients. Epilepsy Res 1993;15:247–52. 18. Cascino GD, Trenerry MR, So EL, et al. Routine EEG and temporal lobe epilepsy: Relation to long-term EEG monitoring, quantitative MRI, and operative outcome. Epilepsia 1996;37:651– 6. 19. Trenerry MR, Jack CR Jr, Cascino GD, et al. MRI hippocampal volumes and memory before and after temporal lobectomy. Neurology 1993;33:1800 –5. 20. Cascino GD, Epilepsy and brain tumors: Implications for treatment. Epilepsia 1990;31:S37–S44. 21. Britton J, Cascino GD, Sharbrough FW, Kelly PJ. Low-grade glial neoplasms and intractable partial epilepsy: Efficacy of surgical treatment. Epilepsia 1994;35:1130 –5. 22. Dodick D, Cascino GD, Meyer FB. Vascular malformations and intractable epilepsy: Outcome after surgical treatment. Mayo Clin Proc 1994;69:741–5. 23. Cascino GD, Kelly PJ, Sharbrough FW, Hirschorn KA. Longterm follow-up of stereotactic lesionectomy in partial epilepsy: Predictive factors and electroencephalographic results. Epilepsia 1992;33:659 – 44.
24. Fish DR, Andermann F, Olivier A, et al. Complex partial seizures and posterior temporal or extratemporal lesions: surgical strategies. Neurology 1991;41:1718 – 84. 25. Kuzniecky R, Murro A, King D, et al. Magnetic resonance imaging in childhood intractable partial epilepsies: Pathological correlation. Neurology 1993;43:681–7. 26. Palmini A, Andermann A, Olivier A, et al. Focal neuronal migrational disorders and intractable partial epilepsy: A study of 30 patients. Ann Neurol 1991;30:741–9. 27. Kuzniecky RI. Magnetic resonance imaging in cerebral developmental malformations and epilepsy. In: Cascino GD, Jack CR Jr, eds. Neuroimaging in epilepsy: Principles and practice. Boston: Butterworth-Heinemann, 1996:51– 63. 28. O’Brien TJ, So EL, Mullan BP, et al. Subtraction ictal SPECT co-registered to MRI improves clinical usefulness of SPECT in localizing the surgical seizure focus. Neurology 1998;50:445– 454. 29. O’Brien TJ, So EL, Mullan BP, et al. Extent of resection of the ictal subtraction SPECT focus is an important determinant of epilepsy surgery outcome. Epilepsia 1996; 37(Suppl 5):S182. 30. So EL, O’Brien TJ, Mullan BP, et al. Seizure localization in intractable partial epilepsy by subtraction SPECT co-registered to MRI: A study of 47 consecutive patients. Epilepsia 1996;37(Suppl 5):S124.
J EPILEPSY, VOL. 11, NO. 3, 1998 129