Professional Documents
Culture Documents
Abstract—Background: Although detection of concordant lesions on MRI significantly improves postsurgical outcomes in
focal epilepsy (FE), many conventional MR studies remain negative. The authors evaluated the role of phased array
surface coil studies performed at 3 Tesla (3T PA MRI). Methods: Forty patients with medically intractable focal epilepsies
were prospectively imaged with 3T PA-MRI including high matrix TSE T2, fluid attenuated inversion recovery, and
magnetization prepared rapid gradient echo. All patients were considered candidates for epilepsy surgery. 3T PA-MRIs
were reviewed by a neuroradiologist experienced in epilepsy imaging with access to clinical information. Findings were
compared to reports of prior standard 1.5T MRI epilepsy studies performed at tertiary care centers. Results: Experienced,
unblinded review of 3T PA-MRI studies yielded additional diagnostic information in 48% (19/40) compared to routine
clinical reads at 1.5T. In 37.5% (15/40), this additional information motivated a change in clinical management. In the
subgroup of patients with prior 1.5T MRIs interpreted as normal, 3T PA-MRI resulted in the detection of a new lesion in
65% (15/23). In the subgroup of 15 patients with known lesions, 3T PA-MRI better defined the lesion in 33% (5/15).
Conclusion: Phased array surface coil studies performed at 3 Tesla read by an experienced unblinded neuroradiologist can
improve the presurgical evaluation of patients with focal epilepsy when compared to routine clinical 1.5T studies read at
tertiary care centers.
NEUROLOGY 2005;65:1026–1031
Approximately 60% of all patients with epilepsy have with the EEG focus and can be completely resected.8
focal epilepsy (FE) syndromes, a total of 0.4% of the Therefore increasing the sensitivity of MRI is impor-
population in industrialized countries. Approxi- tant to the diagnostic evaluation of epilepsies.
mately 15% of these patients are not adequately con- In the hands of reviewers experienced in reading
trolled with anticonvulsant drugs and may benefit MRI studies of epilepsy patients given the probable
from epilepsy surgery.1,2 MRI is an essential part of lobar location of seizure onset, high resolution
the diagnostic evaluation of poorly controlled pa- phased array images at 1.5T increased the lesion
tients since the detection of a lesion within the detection rate by up to 64% and allowed better deter-
epileptogenic region dramatically increases the proba- mination of lesion type when compared to routine
bility of a seizure-free postsurgical outcome.3-5 In tem- high resolution head coil images read by the same
poral lobe epilepsy, the probability of a seizure free reviewers.9 One limitation of this study was the
outcome is 82% with concordant lesions vs 56% for hardware limitation of a four-element array, which
patients with an unremarkable MRI.6 In frontal lobe resulted in only limited spatial coverage. The next
epilepsy, the probability of an excellent outcome is 72% logical steps in improving MR quality are to increase
with a concordant lesion vs 41% when no abnormality phased array coverage by increasing the number of
is detected.7 The best postsurgical outcome is ex- elements in the array and to further increase SNR
pected when the MRI-visible lesion is consistent by imaging at 3T.
From A.A. Martinos Center for Biomedical Imaging (Drs. Knake, Triantafyllou, Wald, Wiggins, Stufflebeam, Shiraishi, Dale, Halgren, and Grant, and G.P.
Kirk and M.T. Foley), Massachusetts General Hospital, Harvard Medical School, Charlestown; Department of Neurology (Dr. Knake), Philipps-University,
Marburg, Germany; The National Center for Epilepsy (Dr. Larsson), Sandvika, Norway; and Department of Radiology (Dr. Grant), Massachusetts General
Hospital, Boston.
Supported in part by the National Center for Research Resources (P41RR14075), the Mental Illness and Neuroscience Discovery (MIND) Institute, the
Föderverein Neurologie, University of Marburg, Germany, the GlaxoSmithKline Grant for Clinical Epileptology, and the Professor Dr. Adolf Schmidtmann-
Stiftung, Germany.
Disclosure: L.L. Wald, PhD, works as a scientific advisor for Siemens. The authors report no conflicts of interest.
Received October 26, 2004. Accepted in final form June 20, 2005.
Address correspondence and reprint requests to Dr. P. Ellen Grant, Director, Pediatric Neuroradiology, Department of Radiology, Neuroradiology Section,
Gray 2, B285, 55 Fruit Street, Boston, MA 02114-2696; e-mail: ellen@nmr.mgh.harvard.edu
Patient
no./age, y Epilepsy 3T PA MRI CIM Surgery PSO Histology
CIM ⫽ change in clinical management based on new MRI findings; PSO ⫽ postsurgical outcome; FLE ⫽ frontal lobe epilepsy; CD ⫽
cortical dysplasia; VNS ⫽ vagal nerve stimulator; iEEG ⫽ invasive intracranial EEG; SF ⫽ seizure free; TLE ⫽ temporal lobe epilepsy;
OLE ⫽ occipital lobe epilepsy; HH ⫽ hypothalamic hamartoma; PNH ⫽ periventricular nodular heterotopia; HS ⫽ hippocampal sclero-
sis; CVA ⫽ cavernous venous angioma; UC ⫽ unchanged.
uted relevant clinical information in 5/15 with a FCD diag- identified on 3T PA MRI. Neither has proceeded to inva-
nosed in addition to the previously noted developmental sive monitoring or surgical resection at this time and
venous anomaly,1 a previously noted region of white mat- therefore the significance of these 3T PA MRI findings is
ter T2 hyperintensity determined to be a FCD1 (figure 2), unclear.
and a larger extent to the previously noted lesions thought
to be candidates for the epileptogenic lesion.3 Two patients
with a larger extent identified on 3T PA MRI have under- Discussion. 3T PA MRI interpreted by an experi-
gone surgery: one with hemimegalencephaly (Patient 30) enced unblinded neuroradiologist had an important
and one with mesial temporal sclerosis (MTS) (Patient 31). clinical impact. Experienced unblinded review of 3T
A topectomy was performed in the hemimegalencephaly PA MRI studies resulted in detection of new lesions
patient with a significant reduction in seizure frequency in 65% of previously MRI-negative patients and
since surgery 19 months ago (Engel class IIB). The histol- added additional information in 50% with temporal
ogy was nonspecific. In Patient 31, 3T PA MRI had shown lobe epilepsy and 40% with frontal lobe epilepsy. The
extensive left MTS extending beyond the suspected proba- 3T PA MRI results allowed previously planned inva-
ble left hippocampal sclerosis. In this patient, a selective sive EEG evaluations to be avoided in 10% of pa-
left amygdalohippocampectomy was performed. MTS was tients. Overall, the clinical management was
confirmed by the histopathology and the patient has been changed in 38% of patients with FE.
seizure free for 12 months. In those two patients 3T PA
Interestingly, 72% (11/15) of patients with normal
MRI is likely to improve the postsurgical outcome as the
1.5T studies but lesions detected on 3T PA MRI had
resection boundaries were tailored to the better defined
malformations of cortical development, 73% (8/11) of
lesion. Future studies need to confirm if 3T PA MRI can
improve the lesion definition and postsurgical outcome in
which were focal cortical dysplasias. These results
patients showing a lesion at 1.5T MRI. are similar to a previous study comparing phased
Group 4: Normal 1.5T head coil study, indeterminate 3T array (PA) to standard head coil imaging where 87%
PA MRI study (2 patients). In two patients with previ- (7/8) of newly detected lesions were malformations of
ously normal head coil MRIs, the 3T surface-coil study cortical development. These lesions can be difficult to
raised the suspicion of a focal cortical lesion. One patient detect on MRI, especially if they are small or result
had right temporal lobe epilepsy. In this patient, a possible in subtle signal abnormalities.12 The improved signal
focal cortical dysplasia was detected in the right inferior to noise of the 3T PA MRI studies resulted in better
frontal gyrus. The other patient had left frontal lobe epi- definition of the gray white junction and a more uni-
lepsy. A probable left frontal transmantle dysplasia was form signal intensity in normal cortex, enabling su-
1028 NEUROLOGY 65 October (1 of 2) 2005
Table 2 Electrophysiologic, clinical, imaging, and pathologic findings in Group 3: 15 patients with abnormal 1.5T head coil study and
abnormal phased array surface coil studies performed at 3 Tesla (3T PA MRI)
Patient
no./age, y SCS Epilepsy 1.5 T MRI 3T PA MRI CIM Surgery PSO Histology
SCS ⫽ success surface coil study; CIM ⫽ change in clinical management based on new MRI findings; PSO ⫽ postsurgical outcome;
OLE ⫽ occipital lobe epilepsy; CVA ⫽ cavernous venous angioma; OL ⫽ occipital lobe; TLE ⫽ temporal lobe epilepsy; PL ⫽ parietal
lobe; DVA ⫽ developmental venous anomaly; TL ⫽ temporal lobe; CD ⫽ cortical dysplasia; SWS ⫽ Sturge Weber Syndrome; MF ⫽
multifocal epilepsy; TS ⫽ tuberous sclerosis; post ATL ⫽ after anterior temporal lobectomy; SSR ⫽ significant reduction in seizure fre-
quency; HS ⫽ hippocampal sclerosis; PLE ⫽ parietal lobe epilepsy; SF ⫽ seizure free; FLE ⫽ frontal lobe epilepsy; PNH ⫽ periven-
tricular nodular heterotopia; CC ⫽ corpus callosum.
perior detection of focal disruptions of these features toring and have remained seizure free to date, albeit
which are hallmarks of focal cortical dysplasias. Pa- with limited follow-up (see table 1). In all patients,
tients with focal cortical dysplasias can benefit by the results had an impact on the surgical procedure
detection of the lesion as surgical success is associ- and a tailored lesionectomy was performed. The
ated with complete removal of the structural longest postoperative follow-up is currently 16
abnormality13-16 and lesion detection may obviate the months.
need for invasive monitoring. In this study all four Due to the findings of 3T PA-MRI, the clinical
patients with complete resection of the newly de- management was changed in 38% of patients. Typi-
tected focal cortical dysplasia avoided invasive moni- cally the findings affected the decision to perform
epilepsy surgery: in most patients, the detection of a homogeneity at 3T decrease the impact of the theo-
new structural lesion encouraged surgery without retical increase in SNR over 1.5T images. The exact
further invasive monitoring. In all patients who increase in SNR has not been measured formally in
avoided invasive monitoring before surgery, the these patients but we estimate a twofold increase in
newly detected lesion was congruent with the local- SNR when moving from 1.5 to 3T and an additional
ization of the ictal onset zone defined by video-EEG- fourfold increase in SNR at the cortex due to the
monitoring. None of the lesions was located in or phased array receivers.21 Therefore, we estimate a
near eloquent cortex. In all patients, an extended six- to eightfold increase in SNR for 3T PA MRI
electrocorticography was performed during the resec- compared to 1.5T head coil MRI.
tion. In a small number of patients, the newly iden- When the reader has knowledge of the seizure
tified lesion (for example a nodular periventricular semiology there is a risk that lesions outside the lobe
heterotopia) was associated with poor surgical out- of interest may be missed and lesions in the lobe of
comes and therefore the patients were no longer con- interest may be overcalled. Although our preliminary
sidered good surgical candidates. postsurgical follow-up is good when the detected le-
Increasing the sensitivity of MRI studies is crucial sion was completely resected suggesting that no
in the presurgical evaluation of medically refractory other epileptogenic lesions are present and patho-
focal epilepsies where the patient’s MRI studies have logic confirmation was obtained in all 11 cases that
been interpreted as normal. Increased sensitivity can went to surgery, further follow-up and larger num-
be obtained when MR images are reviewed by a neu- bers of patients are necessary to determine the false
roradiologist experienced in reading MRI studies of negative rate.
patients with epilepsy with access to clinical infor- The impact of epilepsy imaging experience is diffi-
mation.17 Further increases in sensitivity can be cult to quantify, especially when comparing a reader
achieved by increasing the quality of the MRI stud- more invested in the imaging outcome to a number of
ies. The first improvement in MRI quality at 1.5T readers in clinical practice who are under pressure to
was the development of imaging sequences capable read a large number of studies in a short period of
of providing higher spatial resolution images in a time. A previous study comparing expert and non-
reasonable amount of time compared to standard expert review of routine head coil 1.5T MRI studies
spin echo sequences.18 The next improvement in MRI showed that the sensitivity of the experts was 50%
quality at 1.5T was the use of phased array coils compared to 39% for non-experts. The expert sensi-
instead of a quadrature head coil. This resulted in tivity increased to 91% when high resolution head
signal to noise improvements of three- to fourfold at coil 1.5T MRI epilepsy studies were reviewed.17 In
the cortex.19 The latest improvement in MRI quality this prior study, the experts were also provided in-
we describe here results from increasing the static formation on seizure semiology. In our study, not all
field strength from 1.5T to 3T as well as the use of patients went to surgery and therefore statistical
phased array coils. These advances result in im- sensitivity and specificity could not be calculated.
proved signal to noise which can be used to decrease However, outside reports of high resolution head coil
scan time, increase spatial resolution, and improve 1.5T MRI epilepsy studies noted a focal candidate
contrast to noise ratio (CNR). In our study, the gains lesion in 30% (12/40) of patients whereas focal le-
in SNR were used to increase resolution and to im- sions were detected in 73% (29/40) of patients on
prove CNR.20 Although SNR should increase linearly unblinded review of 3T PA MRI studies.
with field strength, the decrease in T1 contrast be- One limitation of this study is the inability to sep-
tween gray and white matter, the increase in suscep- arate the effects of 3T imaging, PA imaging, an un-
tibility effects, and the difficulties with image blinded reader, and a reader experienced in epilepsy
1030 NEUROLOGY 65 October (1 of 2) 2005
imaging. However, this study does reflect the poten- 7. Mosewich RK, So EL, O’Brien TJ, et al. Factors predictive of the out-
come of frontal lobe epilepsy surgery. Epilepsia 2000;41:843–849.
tial diagnostic yield a referring physician can expect 8. Palmini A, Gambardella A, Andermann F, et al. Intrinsic epileptogenic-
if a patient with FE is referred for 3T PA-MRI eval- ity of human dysplastic cortex as suggested by corticography and surgi-
cal results. Ann Neurol 1995;37:476–487.
uation at a specialized Epilepsy Imaging Center af- 9. Grant PE, Vigneron DB, Barkovich AJ. High-resolution imaging of the
ter routine clinical epilepsy MRI evaluations at brain. Magn Reson Imaging Clin N Am 1998;6:139–154.
tertiary care centers. 10. Triantafyllou C, Dale AM, Fischl B, Knake S, Wald LL. Optimized B1
inhomogeneity correction for high field magnetic resonance imaging.
Neuroimage 2003;19:S707.
Acknowledgment 11. Wald LL, Carvajal L, Moyher SE, et al. Phased array detectors and an
automated intensity-correction algorithm for high-resolution MR imag-
The authors thank Thomas Benner, Andre van der Kouwe, and ing of the human brain. Magn Reson Med 1995;34:433–439.
John Pitts for their help with sequence optimization. They also 12. Tassi L, Colombo N, Garbelli R, et al. Focal cortical dysplasia: neuro-
thank Dr. A. Golja, MD, Department of Radiology, Childrens Hos- pathological subtypes, EEG, neuroimaging and surgical outcome. Brain
pital Boston, and Dr. P. Due Tønnessen, Dept. of Radiology, Riks- 2002;125(Pt 8):1719–1732.
hospitalet Oslo, Norway, for providing the 1.5T MRIs and Drs. 13. Bingaman WE. Surgery for focal cortical dysplasia. Neurology 2004;
Bruce Rosen, Greg Sorensen, and Keiko Hara for their support of 62(6 suppl 3):S30–S34.
the epilepsy program. The authors thank their clinical collabora- 14. Palmini A, Andermann F, Olivier A, Tampieri D, Robitaille Y. Focal
neuronal migration disorders and intractable partial epilepsy: results of
tors for patient referrals: Drs. Ann Bergin, Peter Black, Blaise
surgical treatment. Ann Neurol 1991;30:750–757.
Bourgeois, Edward Bromfield, Keith Chiappa, Rees Cosgrove, An- 15. Kral T, Clusmann H, Blumcke I, et al. Outcome of epilepsy surgery in
drew Cole, Barbara Dworetzky, Daniel Hoch, Gregory Holmes, focal cortical dysplasia. J Neurol Neurosurg Psychiatry 2003;74:183–
Joseph Madsen, James Riviello, Donald Schomer, and Elizabeth 188.
Thiele. 16. Cohen-Gadol AA, Ozduman K, Bronen RA, Kim JH, Spencer DD. Long-
term outcome after epilepsy surgery for focal cortical dysplasia. J Neu-
rosurg 2004;101:55–65.
References 17. Von Oertzen J, Urbach H, Jungbluth S, et al. Standard magnetic reso-
1. Rosenow F, Lüders H. Presurgical evaluation of epilepsy. Brain 2001; nance imaging is inadequate for patients with refractory focal epilepsy.
124:1683–1700. J Neurol Neurosurg Psychiatry 2002;73:643–647.
2. Hauser A. Incidence and prevalence. In: Engel J Jr, Pedley T, eds. 18. Barkovich AJ, Rowley HA, Andermann F. MR in partial epilepsy: value
Epilepsy: a comprehensive textbook. Philadelphia: Lippincott Raven, of high-resolution volumetric techniques. AJNR Am J Neuroradiol
1998;47–57. 1995;16:339–343.
3. Ruggeri P. Magnetic resonance imaging. In: Luders HO, Noachtar S, 19. Lin FH, Chen YJ, Belliveau JW, Wald LL. A wavelet-based approxima-
eds. Epileptic seizures: pathophysiology and clinical semiology. London: tion of surface coil sensitivity profiles for correction of image intensity
Churchill Livingstone, 2000;85–93. inhomogeneity and parallel imaging reconstruction. Hum Brain Mapp
4. Kuzniecky RI, Knowlton RC. Neuroimaging of epilepsy. Semin Neurol 2003;19:96–111.
2002;22:279–288. 20. Briellmann RS, Pell GS, Wellard RM, Mitchell LA, Abbott DF, Jackson
5. Kuzniecky RI. Neuroimaging in pediatric epilepsy. Epilepsia 1996;37 GD. MR imaging of epilepsy: state of the art at 1.5 T and potential of 3
Suppl 1:S10–S21. T. Epileptic Disord 2003;5:3–20.
6. Kuzniecky R, Burgard S, Faught E, Morawetz R, Bartolucci A. Predic- 21. Wald LL, Carvajal L, Moyher SE, et al. Phased array detectors and an
tive value of magnetic resonance imaging in temporal lobe epilepsy automated intensity-correction algorithm for high-resolution MR imag-
surgery. Arch Neurol 1993;50:65–69. ing of the human brain. Magn Reson Med 1995;34:433–439.
AWARD APPLICATIONS
AAN Awards Application Deadline is November 1, 2005
The deadline for most of the American Academy of Neurology scientific and research awards is November 1. Let your
accomplishments be recognized by your neurology peers and the medical community.
For more information on the awards and how to apply, visit the AAN Web site at www.aan.com/professionals/awards.