neuroimaging in refractory epilepsy. current practice and evolving

$: Neuroimaging in refractory epilepsy. Current practice and evolving$
E

R

Na

Na

b

a

ARRA

KIMREE

1

csacobezbgalaItbDr(

h0

ARTICLE IN PRESSG ModelURR-7082; No. of Pages 10

European Journal of Radiology xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

European Journal of Radiology

j ourna l h om epage: www.elsev ier .com/ locate /e j rad

eview

euroimaging in refractory epilepsy. Current practicend evolving trends

. Ramlia,1, K. Rahmata,∗, K.S. Limb,2, C.T. Tanb,2

Department of Biomedical Imaging, University Malaya Research Imaging Centre, MalaysiaNeurology Unit, Department of Medicine, University Malaya, Kuala Lumpur, Malaysia

r t i c l e i n f o

rticle history:eceived 7 August 2014eceived in revised form 9 March 2015ccepted 21 March 2015

a b s t r a c t

Identification of the epileptogenic zone is of paramount importance in refractory epilepsy as the successof surgical treatment depends on complete resection of the epileptogenic zone. Imaging plays an impor-tant role in the locating and defining anatomic epileptogenic abnormalities in patients with medicallyrefractory epilepsy. The aim of this article is to present an overview of the current MRI sequences used

eywords:maging

agnetic resonance imagingefractory epilepsypileptogenic lesion

in epilepsy imaging with special emphasis of lesion seen in our practices. Optimisation of epilepsy imag-ing protocols are addressed and current trends in functional MRI sequences including MR spectroscopy,diffusion tensor imaging and fusion MR with PET and SPECT are discussed.

© 2015 Published by Elsevier Ireland Ltd.

lectroencephalography

. Introduction

About a 15–30% of patients with epilepsy are refractory to medi-al therapy, and surgery is the most effective method for controllingeizures in this group of individuals [1–3]. Pre-surgical evaluationims to delineate the epileptogenic zone, which is a theoreticaloncept of a cortical area that is indispensible for the generationf epileptic seizure [4]. The extent of epileptogenic zone cannote measured directly but the hypothesis of the localisation of thepileptogenic zone can be generated by localisation of other corticalones including ictal onset zone, irritative zone (the area occupiedy interictal discharges in EEG), epileptogenic lesion, symptomato-enic zone, and functional deficit zone. Neuroimaging is essentialnd mandatory in the pre-surgical work-up of the localisation andateralisation of epileptogenic zone. CT scan of the head has thedvantages of easier access, quicker scanning time, and lower cost.t is thus particularly useful in acute symptomatic seizures and sta-us epilepticus, and when MRI is less accessible. CT is also able toetter demonstrate bone pathology, calcification and haemorrhage.

Please cite this article in press as: N. Ramli, et al., Neuroimaging in refr(2015), http://dx.doi.org/10.1016/j.ejrad.2015.03.024

uring epilepsy pre-surgical evaluation, CT is also useful to co-egister invasive electrodes with the magnetic resonance imagingMRI).

∗ Corresponding author. Tel.: +60 379581973; fax: +60 379581973.E-mail address: katt [email protected] (K. Rahmat).

1 Tel.: +60 379581973; fax: +60 379581973.2 Tel.: +603 79494422; fax: +60 379494422.

ttp://dx.doi.org/10.1016/j.ejrad.2015.03.024720-048X/© 2015 Published by Elsevier Ireland Ltd.

Surgery has become an invaluable treatment modality of refrac-tory epilepsy. The role of neuroimaging is in particular to delineatediscrete lesions amenable to surgical resection, such as mesialtemporal sclerosis, focal cortical dysplasia, or hypothalamic hamar-toma. MRI has become the method of choice due to its superior softtissue contrast, multiplanar imaging capability and lack of beamhardening artifacts. MRI of the brain should be the investigationof choice in children and adults with epilepsy to screen for struc-tural abnormalities as recommended by the National Institute ofHealth and Clinical Excellence (NICE) guidelines [5]. In particular,MRI is indicated in those individuals who develop epilepsy beforethe age of 2 years or in adulthood, suggestion of focal onset basedon history, examination or EEG (unless clear evidence of benignfocal epilepsy) and refractory to anti-epileptic medications.

In this article, we first review the applications of MRI in refrac-tory epilepsy with use of conventional imaging protocols withregards to the common cortical abnormalities associated withrefractory epilepsies that are surgically remediable. The second partof this review focuses on the role of EEG and integrated MRI andEEG. The final part looks at other neuroimaging modalities includ-ing CT and nuclear imaging techniques.

2. Conventional MRI imaging protocols in epilepsy

actory epilepsy. Current practice and evolving trends, Eur J Radiol

The recommended epilepsy protocol MRI at 1.5 T or 3.0 Tincludes the entire brain from nasion to inion, T1-weighted mag-netisation prepared rapid gradient echo (MPRAGE), or spoiledgradient recalled (FSPGR) Images 1.5-mm slice thickness with no

dx.doi.org/10.1016/j.ejrad.2015.03.024

dx.doi.org/10.1016/j.ejrad.2015.03.024

http://www.sciencedirect.com/science/journal/0720048X

http://www.elsevier.com/locate/ejrad

mailto:[email protected]

dx.doi.org/10.1016/j.ejrad.2015.03.024

IN PRESSG ModelE

2 nal of Radiology xxx (2015) xxx–xxx

istamchloihotfi

2

dia3swast(wottd1a

2

i(wiihtwnThoe

tfi63Td

“iyest

Table 1Standard brain Epilepsy protocol on 3T MRI at our institution.

Sequence TR/TE Plane Time (min) Slice thickness(mm)

T1 FSPGR 6.7/1.8 Coronal 3.5 1.2 (0.6overlap)

or BRAVO(optional) 7.4/2.9 Coronal 4:0 0.8–1.0FLAIR 8000/120 TL Coronal 7:0 3.0T2 FSE 3760/102 Axial 2:2 5.0T2 FRFSE 7400/102 TL Coronal 3:2 3.0FSEIR 8000/50 TL Coronal 6:0 3.0GE 655/20 Coronal 2.19 5.0

ARTICLEURR-7082; No. of Pages 10

N. Ramli et al. / European Jour

ntervening gap obtained in the coronal oblique plane (if TLE isuspected), coronal, and axial FLAIR sequences with 2–3-mm slicehickness and 0–1-mm interslice gap. These images are acquired as

3D volume, ‘thereby allowing post-processing to correct for headisalignment and for reformatting images into multiple planes to

onfirm a subtle malformation of cortical development [6–9]. Itas been reported that the overall specificity for finding of focal

esion in standard MRI was 22%, if re-assessed by experts 40%, andf dedicated MRI epilepsy protocol is 89%. The use of standard MRIn epilepsy should be limited to exclusion of abnormalities, whichave to be treated irrespective of seizure considerations at the onsetf the disorder like stroke or malignant brain tumours [10]. In ordero optimise a dedicated epilepsy MRI imaging protocol; increasingeld strength, hard ware and sequences contributions are required.

.1. The role of MRI field strength in epilepsy imaging

High field 3 T when compared to the 1.5 T is found to increaseetection of new lesions, providing additional relevant clinical

nformation and detection of lesion in 65% of cases referred with normal routine epilepsy 1.5 T MRI studies [11]. The high field

T with the presence of strong gradients allows the use of 3Dequences of greater signal-to-noise ratio than on the 1.5 T MRI,hich are very useful for small lesions with an advantage of reduced

cquisition time. A recent study on focal epilepsy found that 3 Tcans offer better performance than 1.5 T scans in terms of bet-er characterisation of atrophy and gliosis. Focal cortical dysplasiaFCD) was detected in 11% of patients using 3 T, majority of whomere tested to be negative on 1.5 T MR [12]. A study on the detection

f transmantle sign in type 2 focal cortical dysplasia (FCD2) showedhat 3 T MRI allowed a high spatial resolution in a given acquisitionime with isotropic millimetric voxels. The authors stated that theetection and characterisation of FCD2 was better at 3 T than at.5 T with similar head coils and acquisition time, owing to greaterbility at 3 T to detect the transmantle sign [13].

.2. Hardware requirement in epilepsy imaging

The hard ware required, is a phased array (PA) surface coilnstead of a quadrature head-coil to increase signal to noise ratioSNR) dependence. PA improves SNR up to fivefold in the cortexhich inturns results in an increased lesion detection and diagnosis

n 64% of patients with focal epilepsy [14]. PA imaging at 3 T signif-cantly improves image quality compared with that in routine 1.5 Tead coil studies that will also significantly increase lesion detec-ion in patients with focal epilepsy. In previous study on childrenith intractable epilepsy, high-resolution MRI identified lesionsot detected by standard MRI in more than half the children (56%).echnical advances such as four-coil phased surface array MRI canelp identify and better delineate lesions, improving the diagnosisf patients who are candidates for surgical treatment of refractorypilepsy [15]

Increasing the number of channels within the coil has the effecto provide improved signal-to-noise ratio (SNR) in the peripheryeld of view as well as providing accelerated imaging. For example,4-channel array provides similar SNR in the brain centre as the2-channel array but a 1.3-fold more SNR in the brain cortex [16].his plays an important role in optimizing grey and white matterifferentiation on epilepsy imaging.

High resolution MRI including thin coronal slices, in addition to adynamic” analysis in a workstation with MPR allows a significativemprovement in lesion detection compared to the traditional anal-


sis with radiographic films (94% versus 80%). Patients with focalpilepsy and “normal” MRI need to be investigated further with thinlices and post-processing techniques such as volume acquisitionshat allow adequate multi-planar re-slicing [17]. Semi-automated

FSPGR, Fast Spoiled Gradient Echo; BRAVO, brain volume 3D isotropic FSPGR; FLAIR,fluid-attenuated inversion recovery; FSEIR, fast spin echo inversion recovery GE,gradient echo; TL, temporal lobe angulation.

image-analysis techniques have the potential to improve lesiondetection, assess lesion burden more accurately, characterise cor-tical abnormalities, and determine the location and extent ofassociated cortical and deep grey nuclei involvement. [11]

2.3. Sequence contribution

At our institution routine scanning protocol for a patientwith epilepsy includes: T1 Volumetric 3D spoiled gradient echosequence, T2 W Axial fast spin echo (FSE), Coronal white matterinversion recovery (WMIR), Coronal T2 fluid attenuation inversionrecovery (FLAIR), coronal gradient echo (GE) and high resolutionT2 W FSE (Table 1) 2-plane post-contrast images and spectroscopyare required when there is a need to evaluate for suspected condi-tions such as neoplastic, inflammatory or infectious process. Thesesequences are used with three main aims: (1) to optimise grey andwhite matter differentiation, (2) delineate detail of small structuresand volumetric assessment, and (3) enhance lesion conspicuity.The total scan time for routine epilepsy protocol is approximately30–35 min (Table 1)

2.3.1. Optimisation of grey and white matter differentiationOptimisation in the grey and white matter differentiation assists

in localisation and identification of the various types of grey matterheterotopia, delineation of malformation of cortical development[6] and mesial temporal sclerosis (MTS) [18]. MCD has been increas-ingly recognised as a cause for uncontrolled epilepsy, with theimprovement and utilisation of high-resolution MRI techniques.MCD contributes to 25–45% of medically refractory childhoodepilepsy [19]. The sequence for optimisation of grey and whitematter differentiation is high resolution fast inversion recovery T2-weighted sequence, which is done in near coronal oblique withslice thickness of 3 mm, slice interval of 0.5 mm and scan time ofapproximately 6 min 9 s. This sequence utilises T2-based sequencewith modification of inversion time that will increase the con-trast between grey and white matter (Figs. 1 and 2). T2W Imagesare then inverted at the review console to appear like a T1W. Itis also referred to as white matter inversion recovery (WMIR).There is improved sensitivity of high-resolution protocols com-pared to standard MRI as a result of improved signal to noise ratioand superior anatomical detail [20]. New high-resolution sequencesuch as 3D double inversion recovery (DIR) has been shown tobe more sensitive than FLAIR and T2W in the detection of seizurelaterality in temporal lobe epilepsy [21]. Overall, the largest con-secutive analysis investigating 2000 patients referred for MRI aftera seizure reported epilepsy related abnormalities in roughly 20% ofthe cases. In patients with refractory epilepsy, structural lesions can


be depicted in up to 82–86% of imaging studies by visual inspection[22].

Volumetric 3D spoiled gradient echo sequence (SPGR) orFast Spoiled Gradient Echo (FSPGR) should be performed using

dx.doi.org/10.1016/j.ejrad.2015.03.024

ARTICLE IN PRESSG ModelEURR-7082; No. of Pages 10

N. Ramli et al. / European Journal of Radiology xxx (2015) xxx–xxx 3

Fig. 1. Complex partial seizures in an 18-year-old. Coronal MRI in (a) high resolution fast inversion recovery, (b) FSPGR, (c) and T2-weighted FSE showing multipleperiventricular and subependymal nodules (arrow) at the right lateral ventricle, isointense to cortical grey matter on all sequences consistent with periventricular nodularheterotopia.

2W showing subependymal heterotopia (short arrows) and pachygyria (asterisks).

cncairFdoawduvspc

2a

twipp

Fig. 2. Coronal high-resolution fast inversion recovery (a) T1W and (b) Axial T

ontinuous coronal isotropic acquisition of 1.0–1.2 mm slice thick-ess, without slice interval and perpendicular to the corpusallosum. It provides excellent T1-weighted contrast between greynd while matter and helps to detect subtle cortical dysplasia andnternal structure of hippocampus in patients with mesial tempo-al sclerosis (MTS) [14,23,24]. Scan time is approximately 3–4 min.SPGR allows visualisation of subtle abnormalities that are notemonstrated on T2-weighted sequence and hence is also capablef demonstrating abnormality like cortical dysplasia (Fig. 3). Thiscquisition also utilises the parameters that will augment the grey-hite matter differentiation that in turn improves the anatomicaletails [25–28]. Another sequence newly available is the brain vol-me 3D T1 Bravo sequence (GE Healthcare), which is an enhancedersion of FSPGR. Bravo deploys a 3D IR-prepared FSPGR acqui-ition to produce isotropic T1W volumes that is compatible witharallel imaging, generates images with improved signal to noiseharacteristics and showing less susceptibility artifacts (Fig. 4).

.3.2. Optimizing detail of small structures and volumetricssessment

The sequence to look at the detail of small structures such ashe hippocampal formations is high resolution near coronal T2-

Please cite this article in press as: N. Ramli, et al., Neuroimaging in refractory epilepsy. Current practice and evolving trends, Eur J Radiol(2015), http://dx.doi.org/10.1016/j.ejrad.2015.03.024

eighted FSE acquisition. This sequence is useful in MTS, whichs characterised by selective neuronal loss and gliosis of the hip-ocampus with resultant architectural abnormality of the neuronalathway involving the amygdala, fornix and mammillary body.

Fig. 3. MRI in coronal high resolution 3D FSPGR (Fast Spoiled Gradient Echo)reveals cortical grey matter thickening associated with indistinct grey white mattermargination at the peri-Sylvian right insular cortex (arrow head) compatible withType 2 focal cortical dysplasia.

dx.doi.org/10.1016/j.ejrad.2015.03.024


4 N. Ramli et al. / European Journal of Radiology xxx (2015) xxx–xxx

F 1 mmt y in th

Mswnac

aTavstamaQmmDdmswt

F(T

ig. 4. Comparison between a) BRAVO and b) FSPGR with similar slice thickness ofhe cerebral lobes. The abnormal cortical laminations are displayed in greater clarit

TS classically demonstrates T2-weighted and FLAIR hyperinten-ity. Using 3 T MRI, the T2-weighted fast spin echo (FSE) sequenceith 512 × 384 matrixes and 3–4 mm section thickness done inear coronal plane oblique and scan time of approximately 3 minnd 36 s provides exquisite details of the hippocampal high signalhanges (Fig. 5).

Optimisation in the visualisation of the detailed structures ornatomy of small structures is performed with the use of thin slices.his is again performed using the FSPGR/BRAVO sequences, whichllows multi-planar reconstructions when acquired in isotropicoxel. It provides better anatomical localisation and the 3D acqui-ition datasets can been used for volumetric measurements ofhe hippocampal formations (Fig. 6). Hippocampal volume loss ortrophy can be determined by subjective visual estimation or byeans of objective volume measurement through manual, semi-

utomated or automated MRI computer segmentation methods.uantitative hippocampal volumetric measurement is known to beore sensitive and reliable than visual inspection alone in deter-ining the size affected in various neurological disorders [29].espite the time consuming post processing technique involved inelineating the hippocampal borders for manual or semiautomatic


ethods of measurement, the objective yield is very useful in theetting of less than experienced interpreting reader conditions asell as in presence of bilateral hippocampal atrophy. Furthermore

he measurement is reproducible and reliable for interventional

ig. 5. Left mesial temporal sclerosis with volume loss of the parahippocampal gyrusarrow) as well as the sclerosis of the cornu amonis (*) depicted on high resolution2-weighted coronal view.

. This case of lissencephaly demonstrated with symmetrical loss of sulcation of alle BRAVO sequence.

purposes and future reference. The most important pitfall in quan-titative volumetric and structural neuroimaging is the presence ofpartial volume effects (PVE). Thus, PVE has to be taken into accountnot only during the quantification, but also during possible pre-processing steps of data like fMRI, DTI, or perfusion images [30].

2.3.3. Optimising lesion conspicuityThe FLAIR sequence is helpful to enhance lesion conspicu-

ity especially those located in the subcortical and subependymalregion. Abnormal sulcation and associated abnormal white mattersignal alteration usually occurs in type II focal cortical dysplasia(FCD) that may be apparent as transmantle sign on FLAIR and T2weighted images (Fig. 7) [31]. The FLAIR sequence done in nearcoronal plane oblique is performed orthogonal to the long axis ofthe hippocampus (so called “temporal lobe oblique”) with slicethickness of 3–5 mm, slice interval of 0.5 mm and a scan time ofapproximately 3 min. The two common neoplastic developmentalentities are dysembryoplastic neuroepithelial tumor (DNET) andganglioglioma (Fig. 8). Sequences of susceptibility artifacts such asgradient echo (GRE)/T2* or susceptibility weighted imaging [29] areparticularly useful in cases of trauma or vascular malformations,SWI has been used in the further assessment studies of arterialvenous malformations, occult venous disease, multiple sclerosis,trauma, tumors and functional brain imaging. This sequence ishelpful when searching for haemoglobin breakdown products orto look for calcifications in tuberous sclerosis, or Sturge-Weber[32,33].

3. Integration of EEG with imaging

Electroencephalography (EEG) is the most useful diagnostic pro-cedure for epilepsy, and despite the advent of more sophisticatedmethods of imaging structural damage, epilepsy remains as one ofthe clinical problems routinely requiring EEG evaluation. EEG playsan important role in the diagnosis and classification of epilepsy[34]. In EEG, irritative zone is the area in the brain where spikesare generated during the interictal phase. Ictal onset zone deter-mined by EEG is usually but not necessary a subset of the irritativezone. Ictal onset zone is the region of the brain that is able to gen-erate spontaneous after discharges which is adequate to inducea seizure. Although it provides excellent temporal resolution, the


EEG has limited spatial resolution. By the time the seizure focus isdetected on the EEG monitor it may be some distance away from theactual seizure focus. Intracranial EEG monitoring improves tempo-ral and spatial resolution, however for such an invasive procedure it

dx.doi.org/10.1016/j.ejrad.2015.03.024



Fig. 6. (a) Right mesial temporal sclerosis showing atrophy and loss of internal architecture of the hippocampus on volumetric acquired FSPGR in coronal reformat and T2coronal views. (b) Post processed FSPGR/BRAVO image datasets volume measurement of the hippocampal formation using manual tracing method on ITK-SNAP software( e con

sau

waotMeit(n

dfim

open source) revealed decreased volume of the right hippocampus compared to th

uffers from the small field of view. An invasive EEG derivation maylready show a seizure pattern while the surface EEG seems stillndisturbed.

MRI is useful to identify the epileptogenic lesion in epilepsy,hereas interictal EEG is useful for assessing the irritative zone

nd ictal EEG for ictal onset zone. The variations and limitationsf each investigation in localizing the epileptogenic zone needo be recognised. If a potential epileptogenic lesion identified by

RI, it is essential to have a congruent EEG finding to prove thepileptogenicity of the lesion. Not all lesions evidenced by MRIn epileptic patients are epileptogenic. Thus, further investiga-ions including ictal single photon emission computed tomographySPECT), interictal Positron emission tomography (PET) or intracra-ial EEG monitoring might be necessary.

According to early studies, MRI was reported to be able to


emonstrate structural pathology in about 70% of all patients withocal epilepsy [35,36]. However, the MRI imaging was interpretedn relative to other investigations, in which a subtle MRI abnor-

ality might become more apparent when interpreted with the

tralateral side (117 mm3 versus 2094 mm3).

additional localizing information from other investigations suchas EEG. Many studies had reported that a conventional MRI had90% sensitivity and 85% specificity in the diagnosis of patients withhippocampal sclerosis of epilepsy undergoing temporal lobectomy[23–25,37]. However, most published surgical outcomes depend onconcordance of imaging with ictal and interictal EEG [26,28]. 92%of those with more than 90% lateralisation of the interictal epilep-tiform discharges (IEDs) had a good surgical outcome as comparedto 50% with less than 90% lateralisation [27]. Similarly, seizure out-come is also determined by ictal EEG. Satisfactory surgical outcomeis more likely seen in those with ictal EEG recordings localised tothe lesion (83%), as compared to those with non-lateralised seizures(63%) or seizures that propagate contralaterally (46%) [38].

Lesion negative refractory focal epilepsy is a major challenge inthe epilepsy surgery. Normal MRI has been correlated with poor


surgical outcomes [18,28]. However, subtle dysplasia could resultin “imaging negative” focal epilepsy [39]. In patients with an appar-ently normal MRI, identifying the irritative or ictal onset zone in EEGmight guide further exploration of epileptogenic lesion in the MRI.

dx.doi.org/10.1016/j.ejrad.2015.03.024



pictin

Hii

4

w

FcH

Fig. 7. (a–d) Axial T1W FSPGR, T2W, Coronal FLAIR and IR T1.W images de

ere the role of higher strength MRI such as 3 T and other imag-ng modalities such as PET or SPECT fused with MRI images mayncrease lesion detection.


. Nuclear medicine imaging in epilepsy

FDG-PET has a role in the pre-surgical evaluation of patientsith intractable focal epilepsy [40]. PET has been used to detect

ig. 8. Axial T2W FSE, FSPGR and post contrast FSPGR of dysembryoplastic neuroepitheliaystic lesion which demonstrate nodular peripheral enhancement. The patient’s EEG hoer seizures persisted despite resection of the tumour.

g a focal cortical dysplasia at the right parietal lobe with transmantle sign.

the functional deficit zone rather than the epileptogenic zone. IctalSPECT is able to demonstrate the ictal neuronal activation and is anon-invasive marker of ictal onset zone. Both are particularly use-ful in cases with negative MRI [41]. Interictal PET and ictal SPECT


may provide additional information in some patients that aids clin-ical decision-making. PET and SPECT are usually not indicated forthe majority of patients with epilepsy but play an important rolein the surgical candidates. The epileptogenic focus will typically

l tumour (DNET) in the right posterior temporal lobe showing a well defined corticalwever showed discordant left anterior temporal sharp-and-slow-waves complex.

dx.doi.org/10.1016/j.ejrad.2015.03.024



Fig. 9. Left temporal lobe epilepsy in a 30-year-old man with intractable seizures and lateralizing EEG at the left temporal lobe. MRI coronal FSPGR (a) and T2-weighted axialF s). Icth was

mdiMwitecaiebs

pdawtitroilnes

SE (b) showing focal irregular cortex (FCD) at the left lateral temporal lobe (arrowyper perfusion in the cortical and subcortical areas of the left temporal lobe which

anifest as an area of hyperperfusion if radiotracer was injecteduring the ictal stage or hypoperfusion if it was injected inter-

ctally. It has been reported that the correct localisation rate ofRI, interictal PET and ictal SPECT in temporal lobe epilepsyere 64%, 87% and 81%, respectively; and corresponding rates

n non-temporal lobe epilepsy were 57%, 71%, and 64%, respec-ively [42]. Although the diagnostic accuracy is limited to somextent, PET and SPECT and its co-registration with MRI is espe-ially helpful in improving lesion localisation [43–45]. Ictal SPECTnd interictal FDG-PET as non-invasive functional tools can providemportant information in addition to MRI and video-EEG. The pres-nce of concordance between ictal SPECT and PET predicted aetter long-term post-surgical outcome for extratemporal epilep-ies, compared with temporal epilepsies [46,47].

Practical issues do arise with performing ictal injection of radio-harmaceuticals as needed in SPECT. The Tc-99m HMPOA must berawn and ready for rapid injection on seizure onset. The timingccepted as ictal SPECT is injection of the radiopharmaceuticalsithin 30 s of the seizure onset. In a non-specialised epilepsy unit

his would be near impossible and most studies will be likely peri-ctal demonstrating areas involved in seizure propagation ratherhan the seizure focus itself. Coregistered PET or SPECT with MRIepresents another promising technique for presurgical evaluationf epilepsy, since this procedure offers a potential improvementn epileptogenic zone localisation with very little additional work-


oad or time required. In patients undergoing epilepsy surgery withonlesional MRI or discordant multimodal investigations, furthervaluation can be done by subtraction ictal single photon emis-ion computed tomography (SPECT) coregistered to MRI (SISCOM).

al HMPAO SPECT with fused MR images in coronal (c) and axial (d) demonstratedhighly suggestive of the epileptogenic zone (arrows).

Oertzen et al. prospectively evaluated a cohort of patients undergo-ing presurgical evaluation with SISCOM and found it to be a highlyvaluable diagnostic tool to localise the seizure-onset zone in non-lesional MRI and extratemporal epilepsies as well as to generatehypotheses for site of surgery or intracranial elctrode implantation[48].

Fig. 9 illustrates the congruity of multimodal findings betweenepileptogenic lesion on MRI and epileptogenic zone on the ictalSPECT/CT in a case of intractable temporal lobe epilepsy. Video EEGtelemetry showed electrical spikes originating in the left temporallobe. The MRI was initially reported as normal. Following the resultof intense tracer uptake in the left lateral temporal lobe on ictalSPECT/CT, a second look MRI was prompted which led to the findingof subtle cortical dysplasia in the left temporal lobe.

In recent years, integrated PET/MRI technique has been increas-ingly used at tertiary centres in the diagnostic work up of patientswith refractory epilepsy [42,44,49,50]. In their cohort of paedi-atric patients with non-lesional refractory epilepsy, Rubi et al.reported that PET/MRI was as accurate as PET alone in detecting theepileptogenic zones and in addition it was very useful in the guid-ance of a repeat MRI scan [51]. They found that PET/MRI detected43% of cases with subtle-lesions on repeated MRI scans. Salamonet al. used FDG PET/MRI to identify cortical dysplasia in patientswith therapy-resistant epilepsy [52]. Their protocol added valuefor the 33% of patients with nonconcordant EEG and neuroimag-


ing findings who otherwise would have required other tests, suchas intracranial electrodes, to identify the focus. Another advantageof PET/MRI is to provide more precise presurgical planning as theborders of the cortical lesions can be more clearly identified. Thus,

dx.doi.org/10.1016/j.ejrad.2015.03.024



Fig. 10. (a) MR coronal fast spin echo (FSE) T2W in a 41-year-old woman with intractable complex partial seizures reveals atrophic and hyperintense left hippocampalc monsd confir

ftitPdrt[

5

mMsMods

pihahissitmrlOarpmtms

w

omplex consistent with mesial temporal sclerosis. (b) Multivoxel spectroscopy deemonstrate hypoperfusion in the left hippocampus. The EEG video recording had

urther advances in the surgical treatment of patients with refrac-ory epilepsy lie in the development and validation of newermaging techniques and incorporation of these tools into the mul-imodality presurgical evaluation. The newer simultaneous hybridET-MRI acquisition system has been found to minimise patientiscomfort while maximise clinical information and optimiseegistration of both modalities. Dose reduction when comparedo PET/CT is also an advantage in the hybrid MRI/PET system45,53].

. Evolving trends

In patients with epilepsy, functions including sensory andotor, language and memory have been studied with advancedRI techniques. Advanced imaging techniques include proton

pectroscopy (MR spectroscopy), diffusion tensor imaging (DTI),R perfusion, arterial spin labelling (ASL) and fMRI with blood

xygen level-dependent [54] activation. These techniques detectynamic changes in the brain associated with brain functions andeizure focus [55,44].

MR spectroscopy although lacking in spatial resolution, hasroven to be a sensitive measure to detect metabolic dysfunction

n patients with temporal lobe epilepsy, particularly MTS involvingippocampus. In particular reduction of NAA has assisted in later-lisation of MTS in 65–96% of cases [56,57]. Reduction in NAA isypothesised to reflect neuronal loss and dysfunction that occurs

n the hippocampus region (Fig. 10). The 2 main techniques in MRpectroscopy are single voxel spectroscopy (SVS) or multi-voxelpectroscopy (MVS). SVS permits interrogation of brain metabolitesn a single location selected by the operator. The typical imagingimes are 2–8 min, depending on voxel dimensions. SVS is a rapid

ethod for characterising the metabolic information in a 4–8 cm3

egion of interest; however it is hampered by poor spatial reso-ution and does not address the problem of lesion heterogeneity.n the other hand, the multi-voxel spectroscopy (MVS) covers

large volume of interest (VOI) in a single measurement. MVSequires longer acquisition, precisely shimmed MRI equipment androcessing time, which leads to potential artifacts but providesetabolic data from multiple areas within the region of interest,

umor and surrounding tissue [58]. Although MVS is technically


ore demanding, it provides wider anatomical coverage and betterpatial resolution, taking into account lesion heterogeneity [59].

DTI is a novel MR technique, which has evolved from diffusion-eighted imaging by adding diffusion weighting gradients in

trated depressed NAA. (c) Fused MRI inter-ictal HMPAO SPECT in the same patientmed presence of corresponding sharp waves in the left temporal region.

different directions. DTI provides the means to fully categorisethe degree of anisotropy in the scanned volume using multiplediffusion-weighted images. It has been shown to be useful in local-isation of epileptogenic focus, in which increased mean diffusivityand decreased fractional anisotropy were observed at the seizurefocus [60,61]. Visualisation of white matter tracts done with DTI,enable the surgeon to accurately plan and preserve vital tracts whilemaximizing resected affected lesion [62,63].

FMRI using the BOLD technique has been regarded as a methodof non-invasive mapping of the eloquent cortex and used forassessing memory function in temporal lobe [64]. It can helpdetermine the language dominant hemisphere in both epilepsyand non-epilepsy populations and can also act as a predictorof deficits after temporal lobe resection [43,64]. Emerging newsequence such as the arterial spin labelling (ASL) perfusion has beenshown to be highly accurate in detecting the laterality of a lesion.In particular, asymmetry of ASL perfusion is strongly correlatedwith lateralised clinical symptoms in stroke, CBF asymmetries instenotic-occlusive disease; haemodynamic asymmetries producedby lateralised brain tumors and lateralised temporal lobe epilepsy.Using continuous ASL, one study found abnormal asymmetries ofmedial temporal lobe flow in patients with intractable temporallobe epilepsy [65]. Given that ASL offers visualisation of flow valuesand not metabolism, it may represent a better means of visualizinginterictal hypoperfusion in patients with epilepsy. In addition, ASLdoes not require the use of intravenous contrast agents, which isan advantage in young patients and/or in those with renal impair-ment. It is also an advantage for patients in whom standards ictalperfusion imaging cannot be performed, and in those cases whereinterictal examinations may have to be repeated often. This studybelieves that adding brain perfusion investigations to the MRI pro-tocol in patients with intractable epilepsy may have a number ofadvantages [66]. Pulsed ASL may be a promising approach to detec-ting interictal hypoperfusion in TLE. This method has potential as aclinical alternative to H215O PET due to noninvasiveness and easyaccessibility [67].

EEG-correlated functional MRI (EEG-fMRI) is a functional MRItechnique studying the hemodynamic effect corresponding tointerictal epileptiform activity [68]. This technique has been usedto understand the epileptic network in certain epilepsy syndrome


such as absence seizures, reflex epilepsy and genetic generalisedepilepsies [69–71]. In epilepsy surgery, EEG-fMRI was used tounderstand the network of focal epilepsy, the reason for treat-ment failure, and to guide invasive monitoring [72–74]. Recently,

dx.doi.org/10.1016/j.ejrad.2015.03.024

ING ModelE

nal of

ri

uawccppGoartsei

6

utitarboldpnetmu

C

A

tnMt

R

[

[[

[

[

[

[

[

[

[

[

[

[

[

[[

[

[

[

[

[

[

[

[

[

[[

[


N. Ramli et al. / European Jour

esearchers started to study to role of EEG-fMRI in detecting thectal onset zone [75].

Sundram et al., studied patients with temporal lobe epilepsy,sing voxel-based morphometry (VBM) that were applied to greynd white matter anatomy. They demonstrated that participantsith temporal lobe epilepsy and psychosis have marked corti-

al, subcortical and extratemporal grey and white matter deficitsompared with those with temporal lobe epilepsy alone and thusrovide support for the psychosis literature that also shows thisattern of change [76]. Voxel-based morphometry revealed thatM pathology in TLE extends beyond the hippocampus involvingther limbic areas such as the cingulum and the thalamus, as wells extralimbic areas, particularly the frontal lobe. White mattereduction was found only ipsilateral to the seizure focus, includinghe temporopolar, entorhinal, and perirhinal areas. This pattern oftructural changes is suggestive of disconnection involving prefer-ntially frontolimbic pathways in patients with pharmacologicallyntractable TLE [77].

. Conclusions

Neuroimaging is mandatory and crucial in the diagnostic workp of patients with medically refractory epilepsy as well as forherapeutic assessment in search for potential surgical cure. MRIs considered to be the most important neuroradiological inves-igation tool given its robust capability to demonstrate precisenatomical brain structure with high-resolution detail. The cur-ent trend as seen from this review, is that focal lesions influencerain areas beyond the epileptogenic lesion, across ensemblesf functionally and anatomically connected brain areas. This hased to the hypothesis that epilepsy represents fundamentally aisease of neural networks. Recent technical advances in higherformance MRI have greatly enhanced the ability to illustrateeuroanatomy, neural networks, clarify metabolic information andlucidate functions of the brain. In addition to EEG, which charac-erises epilepsy syndromes and nuclear medicine, that characterise

etabolic activity, the fusion of all these information allow greaternderstanding of the effects of epilepsy on the neural networks.

onflict of interest

None.

cknowledgements

The authors gratefully acknowledge financial support fromhe University of Malaya Institutional Research Grants. Grantos: RP008B-13HTM, and PO012/2014A.The authors thank Hanimohd Taha, Lau Chin Yee and Nurfadzillah for their technical assis-

ance.

eferences

[1] D.C. Taylor, Epilepsy as a chronic sickness: remediating its impact, in: J.J. Engle(Ed.), Surgical treatment of the epilepsy, 2nd ed., Raven, 1993.

[2] L.C.V.J. Meiners, G.H. Jansen, C.W. van Veelen, MR contribution in surgery ofepilepsy, Eur Radiol 9 (1999) 493–507.

[3] H.O. Luders, National Institutes of Health Consensus Conference. Surgery forepilepsy, J Am Med Assoc 264 (1990) 729–733.

[4] F. Rosenow, H. Luders, Presurgical evaluation of epilepsy, Brain 124 (Pt 9) (2001)1683–1700.

[5] T. Stokes, S.E. Juarez-Garcia, A. Camosso-Stefinovic, J. Baker, R Clinical Guide-lines and Evidence Review for the Epilepsies: diagnosis and management in


adults and children in primary and secondary care, in: The epilepsies: clinicalpractice guideline, Royal College of General Practitioners, London, 2004.

[6] C.A. Davie, G. Wenning, G. Barker, P. Tofts, B. Kendall, N. Quinn, et al., Differen-tiation of multiple system atrophy from idiopathic Parkinson’s disease usingproton magnetic resonance spectroscopy, Ann Neurol 37 (2) (1995) 204–210.

[

PRESSRadiology xxx (2015) xxx–xxx 9

[7] K.M. Kelly, S.S. Chung, Surgical treatment for refractory epilepsy: review ofpatient evaluation and surgical options, Epilepsy Res Treat 2011 (2011) 303624.

[8] D.J. Mikulis, T.P. Roberts, Neuro MR: protocols, J Magn Reson Imaging 26 (4)(2007) 838–847.

[9] V.R. Vattipally, R.A. Bronen, MR imaging of epilepsy: strategies for successfulinterpretation, Magn Reson Imaging Clin N Am 14 (2) (2006) 225–247.

10] J. Von Oertzen, H. Urbach, S. Jungbluth, M. Kurthen, M. Reuber, G. Fernandez,et al., Standard magnetic resonance imaging is inadequate for patients withrefractory focal epilepsy, J Neurol Neurosurg Psychiatry 73 (6) (2002) 643–647.

11] P.E. Grant, Structural MR imaging, Epilepsia 45 (Suppl. 4) (2004) 4–16.12] M. Toledo, S. Sarria-Estrada, M. Quintana, C. Auger, X. Salas-Puig, E. Santama-

rina, et al., 3 TESLA MR imaging in adults with focal onset epilepsy, Clin NeurolNeurosurg 115 (10) (2013) 2111–2116.

13] C. Mellerio, M.A. Labeyrie, F. Chassoux, P. Roca, O. Alami, M. Plat, et al., 3 T MRIimproves the detection of transmantle sign in type 2 focal cortical dysplasia,Epilepsia 55 (1) (2014) 117–122.

14] P.E.B.A. Grant, L.L. Waid, W. Dillon, K. Laxer, D. Vigneron, High resolutiomsurface-coil MR of cortical lesions in medically refractory epilepsy: a prospec-tive study, AJNR Am J Neuroradiol 18 (1997) 291–301.

15] M. Goyal, B.A. Bangert, J.S. Lewin, M.L. Cohen, S. Robinson, High-resolution MRIenhances identification of lesions amenable to surgical therapy in children withintractable epilepsy, Epilepsia 45 (8) (2004) 954–959.

16] B. Keil, J.N. Blau, S. Biber, P. Hoecht, V. Tountcheva, K. Setsompop, et al., A 64-channel 3 T array coil for accelerated brain MRI, Magn Reson Med 70 (1) (2013)248–258.

17] S.L. Santos, E. Ghizoni, M. Li Li, F. Cendes, Dynamic assessment of high-resolution MRI with multi-planar reconstruction increases the yield of lesiondetection in patients with partial epilepsy, J Epilepsy Clin Neurophysiol 11 (3)(2005) 111–116.

18] K.M.T. Hashiguchi, N. Murakami, T. Morika, S.O. Suzuki, A. Hiwatashi, T.Yoshiura, et al., Utility of 3-T FLAIR and 3D short tau inversion recovery MRimaging in the preoperative diagnosis of hippocampal sclerosis: direct com-parison with 1.5-T FLAIR MR imaging, Epilepsia 51 (2010) 1820–1828.

19] R.H.H. Guerrini, L. Parmeggiani, Sicca, F, Epilepsy and malformations of thecerebral cortex, in: J. Roger, M. Bureau, C. Dravet (Eds.), Epileptic syndromes ininfancy, childhood and adolescence, 3rd ed., John Libbey, London, UK, 2002.

20] H. Urbach, High-field magnetic resonance imaging for epilepsy, NeuroimagingClin N Am 22 (2) (2012) 173–189, ix–x.

21] E. Morimoto, M. Kanagaki, T. Okada, A. Yamamoto, N. Mori, R. Matsumoto,et al., Anterior temporal lobe white matter abnormal signal (ATLAS) as anindicator of seizure focus laterality in temporal lobe epilepsy: comparison ofdouble inversion recovery, FLAIR and T2W MR imaging, Eur Radiol 23 (1) (2013)3–11.

22] I.J. Craven, P. Griffiths, D. Bhattacharyya, R. Grunewald, T. Hodgson, D. Connolly,et al., 3.0 T MRI of 2000 consecutive patients with localisation-related epilepsy,Br J Radiol 85 (1017) (2012) 1236–1242.

23] C.G.M. Raybound, J. Mancini, N. Girard, Neuroimaging of epilepsy in children.(viii), Magn Reson Imaging Clin N Am 9 (2001) 121–147.

24] P.M.N.I. Ruggieri, MR imaging in epilepsy, Neurol Clin 19 (2001) 477–489.25] A.J.K.R. Barkovich, G.D. Jackson, R. Guerrini, W.B. Dobyns, Classification system

for malformations of cortical development, Neurology (2002) 2168–2178.26] N.S.N. Colombo, C. Raybaud, C. Ozkara, A.J. Barkovich, Imaging of malformations

of cortical development, Epileptic Discord 11 (2009) 194–205.27] C.L.M. Mellerio, F. Chassoux, M.-A. Labeyrie, C. Daumas-Duport, E. Landre, B.

Turak, et al., Optimizing MR imaging detection of type 2 focal cortical dysplasia:best criteria for clinical practice, Am J Neuroradiol 33 (10) (2012).

28] P.D.B. Widdness-Walsh, I. Najm, Neuroimaging of focal cortical dysplasia, JNeuroimaging 16 (2006) 185–196.

29] C.R. Jeukens, Mariëlle C.G. Vlooswijk, H.J. Marian Majoie, Marc CTFM de Krom,Albert P. Aldenkamp, Paul A.M. Hofman, Jacobus F.A. Jansen, Walter H. Backes,Hippocampal MRI volumetry at 3 Tesla: reliability and practical guidance,Investigative Radiol 44 (9) (2009) 509–517.

30] J. Klein, F. Laun, P. Erhard, V. Diehl, H.K. Hahn, On the reliability of quantitativevolumetric and structural neuroimaging, Imaging Decisions MRI 12 (3) (2008)2–10.

31] A.J.K.R. Barkovich, G.D. Jackson, R. Guerrini, W.B. Dobyns, A developmental andgenetic classification for malformations of cortical development, Neurology 65(2005) 1873–1887.

32] K.A. Tong, S. Ashwal, A. Obenaus, J. Nickerson, D. Kido, E. Haacke, et al.,Susceptibility-weighted MR imaging: a review of clinical applications in chil-dren, AJNR Am J Neuroradiol 29 (1) (2008) 9–17.

33] T.A. Huisman, S. Singhi, P.S. Pinto, Non-invasive imaging of intracranial pedi-atric vascular lesions, Childs Nerv Syst 26 (10) (2010) 1275–1295.

34] S.R.J. Noachtar, The role of EEG in epilepsy: a critical review, Epilepsy Behav 15(2009) 22–33.

35] J. CR, Epilepsy: surgery and imaging, Radiology 189 (1993) 635–646.36] R.A. Bronen, Epilepsy: the role of MR imaging, Am J Roentgenol 159 (1992)

1165–1174.37] M.K.K. Seidenberg, J. Parrish, K.G. Kelly, E. Geary, C. Dow, P. Rutecki, et al.,

Ipsilateral and contralateral MRI volumetric abnormalities in chronic unilat-eral temporal lobe epilepsy and their clinical correlates, Epilepsia 46 (420-30)


(2005).38] J.T.S.N. Lerner, J.S. Hauptman, N. Salamon, T.R. Velasco, M. Hemb, J.Y. Wu, et al.,

Assessment and surgical outcomes for mild type I and severe type II corti-cal dysplasia: a critical review and the UCLA experience, Epilepsia 50 (2009)1310–1315.

dx.doi.org/10.1016/j.ejrad.2015.03.024

http://refhub.elsevier.com/S0720-048X(15)00146-1/sbref0005















































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































ING ModelE

1 nal of

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[


0 N. Ramli et al. / European Jour

39] I.J.G.P. Craven, D. Bhattacharyya, P.D. Griffiths, R.A. Grunewald, T. Hodg-son, D.J.A. Connolly, et al., 3.0-tesla MRI of 2000 consecutive patients withlocalisation-elated epilepsy, Br J Radiol 85 (2012) 1236–1242.

40] R. Casse, C.C. Rowe, M. Newton, S.U. Berlangieri, A.M. Scott, Positron emissiontomography and epilepsy, Mol Imaging Biol 4 (5) (2002) 338–351.

41] Y. Chinvarun, Application of PET and SPECT in epilepsy, Neurology Asia 18 (1)(2013) 23–25.

42] N.K.P.J. Logothetis, M. Augath, T. Trinath, A. Oeltermann, Neurophysiologicalinvestigation of the basis of the fMRI signal, Nature 412 (2001) 150–157.

43] N. Bargallo, Functional magnetic resonance imaging: new applications inepilepsy, Eur J Radiol 6 (2008) 19–26.

44] S.Y.T. Sunaert, Clinical applications of functional magnetic resonance imaging,Neuro Imaging Clin N Am 11 (2001) 221–236.

45] M.I. Vargas, M. Becker, V. Garibotto, S. Heinzer, P. Loubeyre, J. Gariani, et al.,Approaches for the optimization of MR protocols in clinical hybrid PET/MRIstudies, MAGMA 26 (1) (2013) 57–69.

46] P.S. Chandra, Gaurang Vaghania, Chandra Sekhar Bal, Madhavi Tripathi, NileshKuruwale, Amit Arora, et al., Role of concordance between ictal-subtractedSPECT and PET in predicting long-term outcomes after epilepsy surgery,Epilepsy Res. 108 (10) (2014) 1782–1789.

47] Y.S. Ding, B.-B. Chen, C. Glielmi, K. Friedman, O. Devinsky, A pilot study inepilepsy patients using simultaneous PET/MR, Am J Nucl Med Mol Imaging4 (5) (2014) 459–470.

48] T.J. Von Oertzen, F. Mormann, H. Urbach, K. Reichmann, R. Koenig, H. Clusmann,H.J. Biersack, C.E. Elger, Prospective use of subtraction ictal SPECT coregisteredto MRI (SISCOM) in presurgical evaluation of epilepsy, Epilepsia 52 (12) (2011)2239–2248.

49] S.Y.S. Bano, V. Chaudhary, U.C. Garga, Neuroimaging in epilepsy, J Pediatr Neu-rosci 6 (2011) 19–26.

50] B. N, Functional magnetic resonance imaging: new applications in epilepsy, EurJ Radiol 6 (2008) 19–26.

51] S.S.X. Rubí, A. Donaire, X. Setoain, N. Bargalló, F. Sanmartí, M. Carreno, et al.,Validation of FDG-PET/MRI coregistration in nonlesional refractory childhoodepilepsy, Epilepsia 52 (2011) 2216–2224.

52] N.K.A. Salmom, S.J. Shaw, J. Kung, J. Koo, S. Koh, J. Wu, et al., FDG-PET/MRI coreg-istration improves detection of cortical dysplasia in patients with epilepsy,Neurology 71 (2008) 1594–1601.

53] V. Garibotto, S. Heinzer, S. Vulliemoz, R. Guignard, M. Wissmeyer, M. Seeck,et al., Clinical applications of hybrid PET/MRI in neuroimaging, Clin Nucl Med38 (1) (2013) e13–e18.

54] W.B.E. Penfield, Somatic motor and sensory representation in the cerebral cor-tex of man as studied by electrical stimulation, Brain 60 (1937) 389–443.

55] N.K. Logothetis, J. Pauls, M. Augath, T. Trinath, A. Oeltermann, Neuro-physiological investigation of the basis of the fMRI signal, Nature 412 (2001)150–157.

56] E.S.P. Achten, P. Boon, D. De Coo, T. Van De Kerckhove, J. De Reuck, et al.,Single-voxel proton MR spectroscopy and positron emission tomography forlateralization of refractory temporal lobe epilepsy, AJNR Am J Neuroradiol 19(1) (1998) 1–8.

57] H.O.N. Aydin, V. Kizilgoz, E. Altin, I.G. Tatar, B. Hekimoglu, Value of proton-MR-spectroscopy in the diagnosis of temporal lobe epilepsy: correlation of


metabolite alterations with electroencephalography, Iran J Radiol 9 (1) (2012)1–11.

58] E.A. Steffen-Smith, et al., Single- and multivoxel proton spectroscopy in pedi-atric patients with diffuse intrinsic pontine glioma, Int J Radiat Oncol Biol Phys84 (3) (2012) 774–779.

[

PRESSRadiology xxx (2015) xxx–xxx

59] V.A. Nagar, J. Ye, M. Xu, W. Ng, T. Yeo, P. Ong, et al., Multivoxel MR spec-troscopic imaging—distinguishing intracranial tumours from non-neoplasticdisease, Ann Acad Med Singapore 36 (5) (2007) 309–313.

60] B.A.M.F. Assaf, K.J. Abou-Khaled, J.M. Williams, M.S. Yazeji, J. Haselgrove, S.H.Faro, et al., Diffusion tensor imaging of the hippocampal formation in temporallobe epilepsy, Am J Neuroradiol 24 (2003) 1857–1862.

61] H.W.P.G. Powell, D.C. Alexander, G. Parker, M. Symms, P. Boulby, C. Wheeler-Kingshott, et al., MR tractography predicts visual field defects followingtemporal lobe resection, Radiology 65 (2005) 596–599.

62] J. Zhang, W. Liu, H. Chen, H. Xia, Z. Zhou, S. Mei, et al., Multimodal neuroimagingin presurgical evaluation of drug-resistant epilepsy, Neuroimage Clin 4 (2014)35–44.

63] N.H. Ulrich, R. Kockro, D. Bellut, C. Amaxopoulou, O. Bozinov, J.K. Burkhardt,et al., Brainstem cavernoma surgery with the support of pre- and postoperativediffusion tensor imaging: initial experiences and clinical course of 23 patients,Neurosurg Rev 37 (3) (2014) 481–491 [discussion 492].

64] L.S.B.B. Medina, C. Dunoyer, B. Bernal, L. Cervantes, M. Rodriguez, E. Pacheco,et al., Seizure disorders: functional MR imaging for diagnostic evaluation andsurgical treatment-prospective study, Radiology 236 (2005) 247–253.

65] G.G. Brown, C. Clark, T.T. Liu, Measurement of cerebral perfusion with arte-rial spin labeling: Part 2. Applications, J Int Neuropsychol Soc 13 (3) (2007)526–538.

66] N. Pendse, et al., Interictal arterial spin-labeling MRI perfusion in intractableepilepsy, J Neuroradiol 37 (1) (2010) 60–63.

67] Y.M. Lim, Y.-W. Cho, S. Shamim, J. Solomon, R. Birn, W.M. Luh, et al., Usefulnessof pulsed arterial spin labeling MR imaging in mesial temporal lobe epilepsy,Epilepsy Res 82 (2–3) (2008) 183–189.

68] A. Al-Asmi, et al., fMRI activation in continuous and spike-triggered EEG-fMRIstudies of epileptic spikes, Epilepsia 44 (10) (2003) 1328–1339.

69] P.W. Carney, et al., The frontal lobe in absence epilepsy: EEG-fMRI findings,Neurology 78 (15) (2012) 1157–1165.

70] M. Sandhya, Rose Dawn Bharath, Rajanikant Panda, S.R. Chandra, NaveenKumar, Lija George, A. Thamodharan, Arun Kumar Gupta, P. Satishchandra,Understanding the pathophysiology of reflex epilepsy using simultaneous EEG-fMRI, Epileptic Disord 16 (1) (2014) 19–29.

71] B. Kay, Jerzy P. Szaflarski, EEG/fMRI contributions to our understanding ofgenetic generalized epilepsies, Epilepsy Behav 34 (2014) 129–135.

72] M. Avesani, S. Giacopuzzi, L.G. Bongiovanni, P. Borelli, R. Cerini, R. Pozzi Mucelli,et al., EEG-fMRI evaluation of patients with mesial temporal lobe sclerosis,Neuroradiol J 27 (1) (2014) 45–54.

73] J. Zhang, Q. Liu, S. Mei, X. Zhang, X. Wang, W. Liu, et al., Presurgical EEG-fMRI in a complex clinical case with seizure recurrence after epilepsy surgery,Neuropsychiatr Dis Treat 9 (2013) 1003–1010.

74] P.J. van Houdt, J.C. de Munck, F.S. Leijten, G.J. Huiskamp, A.J. Colon, P.A. Boon,et al., EEG-fMRI correlation patterns in the presurgical evaluation of focalepilepsy: a comparison with electrocorticographic data and surgical outcomemeasures, Neuroimage 75 (2013) 238–248.

75] S. Tousseyn, P. Dupont, K. Goffin, S. Sunaert, W. Van Paesschen, Sensitivity andspecificity of interictal EEG-fMRI for detecting the ictal onset zone at differentstatistical thresholds, Front Neurol 5 (2014) 131.

76] F. Sundram, M. Cannon, C.P. Doherty, G.J. Barker, M. Fitzsimons, N. Delanty,


et al., Neuroanatomical correlates of psychosis in temporal lobe epilepsy: voxel-based morphometry study, Br J Psychiatry 197 (6) (2010) 482–492.

77] N. Bernasconi, S. Duchesne, A. Janke, J. Lerch, D. Collins, A. Bernasconi, et al.,Whole-brain voxel-based statistical analysis of gray matter and white matterin temporal lobe epilepsy, Neuroimage 23 (2) (2004) 717–723.

dx.doi.org/10.1016/j.ejrad.2015.03.024












































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































































neuroimaging in refractory epilepsy. current practice and evolving

Documents