www.elsevier.com/locate/jocn

Journal of Clinical Neuroscience 14 (2007) 841–849

Clinical study

Contrast-enhanced perfusion and diffusion MRI accuratelylateralize temporal lobe epilepsy: A pilot study

T.J. O’Brien a,b,c,*, E.P. David a, C.J. Kilpatrick a, P. Desmond d, B. Tress d

a Department of Neurology, The Royal Melbourne Hospital, The University of Melbourne, Victoria, Australiab Department of Medicine, The Royal Melbourne Hospital, The University of Melbourne, Parkville, Victoria, 3050, Australia

c Department of Surgery, The Royal Melbourne Hospital, The University of Melbourne, Victoria, Australiad Department of Radiology, The Royal Melbourne Hospital, The University of Melbourne, Victoria, Australia

Received 6 April 2006; accepted 5 July 2006

Abstract

Aims: To undertake a pilot study to assess whether magnetic resonance (MR) contrast-enhanced perfusion imaging (CEPI) and diffu-sion-weighted imaging (DWI) provide lateralizing information in medically refractory temporal lobe epilepsy (TLE),and to compare thisto standard quantitative hippocampal assessments (volumetric measurements and T2 relaxometry).Methods: Ten patients with ‘non-lesional’ TLE and 10 control subjects were studied. Quantification of the relative cerebral blood flow(rCBF) and apparent diffusion coefficient (ADC) was performed for the hippocampal regions. The ratios of the ipsilateral-to-contralat-eral side (to the EEG lateralization) were compared with the side-to-side ratios in the controls.Results: Six patients (60%) had an ADC ratio outside the control range (the larger ADC ipsilateral to the EEG lateralization in all cases).The CBF ratios were outside the control range in all eight patients (100%) in whom CEPI was performed (the lower value ipsilateral tothe EEG lateralization in all cases). The magnitude of the hippocampal volume (HV) ratios showed no significant correlation with themagnitude of the ADC ratios (R = �0.03, p = 0.93) or CBF ratios (R = 0.36, p = 0.39). There was a closer relationship with the T2 relax-ometry ratios, but this was also not significant (R = �0.40, p = 0.32; R = 0.58, p = 0.08).Conclusions: DWI and CEPI show potential as reliable tools for the lateralization of non-lesional TLE. Further studies with larger num-bers are necessary to determine whether these techniques provide independent data to established MR quantitative measures.� 2006 Elsevier Ltd. All rights reserved.

Keywords: Temporal lobe epilepsy; MRI; Diffusion; Perfusion; Hippocampus; Volumetry; T2 relaxometry

1. Introduction

Successful surgery for medically refractory partial epi-lepsy is critically dependent on the accurate localizationof the epileptogenic zone.1 Temporal lobe epilepsy (TLE)is the most common form of partial epilepsy in adults,and is frequently resistant to medical therapy. A discreteneocortical lesion is found to be the underlying cause in20–40% of refractory TLE cases. In ‘non-lesional’ TLE,

0967-5868/$ - see front matter � 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.jocn.2006.07.003

* Corresponding author. Tel.: +61 3 8344 3260.E-mail address: obrientj@unimelb.edu.au (T.J. O’Brien).

mesial temporal sclerosis (MTS) is the most common path-ological finding following temporal lobectomy, and hasbeen shown to be highly predictive of a good post-opera-tive outcome with respect to seizures.2–4 High-resolutionmagnetic resonance imaging (MRI) can reliably detectMTS pre-operatively, based on the typical findings of uni-lateral hippocampal atrophy and associated increased T2signal.5 Quantitative hippocampal assessment methods,hippocampal volumetry and T2 relaxometry, improve thesensitivity of MRI detection of MTS and are predictiveof patients likely to have a good outcome following epi-lepsy surgery.2,6,7 However, not all patients with refractoryTLE have unilateral MTS or a discrete neocortical lesion

842 T.J. O’Brien et al. / Journal of Clinical Neuroscience 14 (2007) 841–849

that is detectable with current MRI techniques, and thepost-surgical outcome in these patients is significantlypoorer.1,3,8 Invasive intracranial EEG studies are usuallyrequired in these patients to localize the epileptogenic zoneprior to epilepsy surgery. Clearly, therefore, other non-invasive measures of dysfunction are required that providepreoperative localization of the epileptogenic zone in thesepatients.

Advances in MRI technology have resulted in a numberof new techniques being introduced into clinical practicewhich image different types of cerebral dysfunction fromthat of the traditional structural imaging. Two importantexamples are dynamic susceptibility contrast-enhancedMR perfusion imaging (CEPI) and diffusion-weightedMR imaging (DWI). DWI can be used to measure theapparent diffusion coefficient (ADC) of brain water, pro-viding anatomically resolved information about the physi-cal state of water in living tissues. It has been shown todetect changes occurring within minutes of acute ischae-mia, which likely reflects the effects of the interruption toenergy metabolism — although the mechanisms for thisare still incompletely understood.9,10 It is also probablethat the functional and structural changes related to re-peated focal seizures may affect the diffusion of brainwater, and therefore DWI may potentially be useful indefining the seizure focus in epilepsy. In support of thishypothesis are some animal studies,11,12 and several rela-tively small clinical studies in patients with TLE.13–19

Contrast-enhanced MR perfusion imaging allows cere-bral maps of relative cerebral blood flow (rCBF) and cere-bral blood volume (rCBV) to be derived from the changesin tissue signal intensity that follow the first pass of anintravenous paramagnetic contrast agent (gadolinium).20,21

It is well established, based on previous blood flow singlephoton emission computerised tomomgraphy (SPECT)and positron emission tomography (PET) studies, thatrCBF is decreased interictally in the epileptogenic temporallobe in many patients with TLE.22,23 However, both inter-ictal blood flow SPECT and PET have proved to have arelatively poor sensitivity and specificity as a clinical toolfor the lateralization of TLE.23–25 Both these modalitiesare hampered by relatively poor spatial resolution, partialvoluming effects in the mesial temporal regions, and diffi-culties with quantitation. MR perfusion imaging, with itsgreater spatial resolution (2–3 mm),26 has the potential tomore accurately and reliably detect temporal hypoperfu-sion in the epileptogenic zone and thereby be a more usefulclinical tool. However, as yet there have been few system-atic studies assessing the value of this test in localizingthe seizure focus in patients with refractory TLE.27,28

This paper reports a pilot study which aims to assesswhether CEPI and DWI reliably lateralize the epilepto-genic temporal lobe in a group of patients with TLE with-out a discrete neocortical lesion, and determine if theinformation provided by these tests is independent of stan-dard visual assessment and quantitative measures (that is,hippocampal volumetry and T2 relaxometry).

2. Methods

2.1. Subjects

Ten adults (9 female, 1 male) with medically refractorynon-lesional TLE (that is, without a discrete structural le-sion detected on MRI, excluding mesial temporal sclerosis),who had been found on pre-surgical non-invasive video-EEG monitoring to have well lateralized ictal discharges(6 left-sided, 4 right-sided), were studied. In addition, 10normal subjects (5 male and 5 female) without a historyof seizures or other neurological disease were studied ascontrols. The age of the two groups was not significantlydifferent (median 35.5 years, range 21–61 vs. median 26.5years, range 21–39, p > 0.05 Mann-Whitney U-test). Sevenof the patients had unequivocal MRI evidence of unilateralhippocampal sclerosis (HS), based on the findings on visualinspection of hippocampal volume loss and/or increasedsignal on T2 weighted images. In all cases the side of theMRI-identified HS was ipsilateral to the side of the ictalEEG seizure lateralization. As part of their pre-surgicalevaluation, all the patients underwent MRI, interictalEEG, video EEG monitoring, PET and neuropsychologyassessment. Seven of the patients have since proceeded toan anterior temporal lobectomy and amydalohippocampec-tomy. The pathological examination demonstrated changesof mesial temporal sclerosis in six patients and non-specificcell loss and gliosis in the other. All patients have had agood outcome with respect to seizures during a post-operative follow-up period ranging from 12–24 months,with four remaining completely seizure-free; two patientshad seizures only related to sudden cessation of their AEDs;and one had rare spontaneous seizures.

The study was approved by the Royal Melbourne Hos-pital Human Ethics and Research Committee, and writteninformed consent was obtained from all patients andcontrols.

2.2. Magnetic resonance imaging acquisition

The patients and controls were scanned on a 1.5-T clin-ical whole-body scanner (Sigma Horizon SR120, GE Med-ical Systems, Milwaukee, WI, USA) using an identicalstandardized protocol including a T1-weighted sagitallocalizer, a coronal T1-weighted whole brain volumetricseries, an oblique-coronal DWI sequence, oblique coronalT2 mapping sequence, and an oblique CEPI sequence.The ‘whole brain’ volumetric series was acquired using afast spoiled gradient echo technique (FSPGR) with a sliceof 1.5 mm thickness, zero interslice gap, at 11/2 ms (TR/TE) pulse sequence, a 22 · 22 cm field of view (FOV),and a 256 · 256 matrix size. The T2 mapping acquisitionconsisted of 12 oblique coronal 5 mm slices with zero gapthrough the hippocampus acquired with conventional spinechoes of TR 4000 ms and TE 30, 60, 120, 180 ms (fourechoes). Field of view was 40 · 20 mm with a 256 · 128matrix.

T.J. O’Brien et al. / Journal of Clinical Neuroscience 14 (2007) 841–849 843

Diffusion-weighted MRI scans were performed using asingle-shot, spin-echo, echo planar imaging (EPI) sequencewith the Stejskal-Tanner diffusion encoding method.29 Slicethickness was 5 mm with zero interslice gap. The number ofslices was set at 12 which included the entire temporal re-gions. Matrix size was 256 · 128 and TR/TE was 10 s/110ms. Diffusion gradient strength was varied to obtain fiveb-values of increasing magnitude from 0 to 800 sec/mm2.The diffusion gradients were applied in six directions. Threeorthogonal diffusion images were used to form the isotropicdiffusion map and the trace of the tensor was calculated toform the apparent diffusion coefficient (ADC) maps.30,31

Contrast-enhanced MR perfusion images were obtainedin eight patients and nine control subjects (in the other 2 pa-tients and 1 control subject technical problems with theinjector pump resulted in this sequence not being obtained).A bolus of gadolinium diethylenetriamine penta-acetic acid(Gd-DTPA) (0.1 mmol/kg) was administered via a largebore catheter in the antecubital fossa by a mechanical injec-tor (Spectris MR injector, MEDRAD, Indianola, PA,USA) at a rate of 5 mL/sec. An EPI gradient echo sequencewas used with a TR/TE of 2000/39 msec and a flip angle of60�. A total of 12 slices were obtained in identical positionsas the DWI images, with a slice thickness of 5 mm, zerointerslice gap, a matrix of 256 · 128 and an FOV of40 · 20 cm. Images were obtained at 40 time points per slicewith a total imaging time of 1 min and 21 sec.

2.3. Visual MRI assessment

The routine MRI images were visually assessed by a sin-gle reviewer (PD) who was blinded to the clinical features orany lateralizing information. The reviewer was asked to as-sess for unilateral atrophy or asymmetry of the volume ofthe hippocampal formation (on the coronal T1 slices), andalso for the presence of relative increased signal in one hip-pocampus compared to the other. These results were thencompared to the results from the DWI and CEPI analysis.

2.4. Diffusion weighted image processing

Post-processing of the DWI images was performed on aUnix-based workstation using customized software devel-

Fig. 1. Demonstrates the site of placement of the regions of interest (ROIs) oveimages which were then applied to the apparent diffusion co-efficient (ADC) m

oped in IDL (Interactive Data Language, Research Sys-tems Inc., Boulder, CO, USA) by a single operator whowas blinded to whether the images were from a patient orcontrol, and to any clinical or lateralizing data. Rawimages were filtered with a 9 · 9 Gaussian kernel(r = 0.5). A region of interest (ROI) was drawn free-handover both hippocampal regions on the raw DWI images,and this ROI was then applied to the ADC maps(Fig. 1). From this, the average ADC value per pixel wascalculated for each side.

2.5. Perfusion weighted image processing

Post-processing and analysis of the perfusion imageswas also performed on an off-line Unix-based workstationby the same blinded operator (QY) with the aid of custom-ized software. The arterial input function, AIF(t), wasdetermined from pixels within the middle cerebral artery(MCA) which was identified visually and confirmed bythe early and high concentration maximum of the contrastbolus. Concentration time curves were then generated fromthe changes in signal intensity over time assuming a linearrelationship between T2* rate change (D R2*) and intravas-cular concentration of Gd-DTPA (Fig. 2).32,33 The concen-tration time curves were then deconvolved with the AIF(t)and fitted with a gamma variate function to correct for theeffects of contrast recirculation on a pixel-by-pixel basis.Maps of rCBF and rCBV were generated from the normal-ized concentration time curves, with rCBF taken from theheight and rCBV from the area under these curves.34,35

ROIs covering the hippocampi were manually drawn ona mid-hippocampal slice of the coronal EPI gradient echobaseline images as for the DWI analysis, and these werethen applied to the rCBF and rCBV maps.

2.6. Hippocampal volume measurements

The hippocampal volumetric measurements were per-formed off-line with the aid of a commercial image analysissoft-ware package (ANALYZE 7.2�). Hippocampal vol-ume measurements were performed on the volumetriccoronal T1-weighted images using a standardized protocolas previously described.36 The images were first magnified

r the mid-hippocampal bodies bilaterally on the coronal diffusion-weightedaps.

Fig. 2. A concentration-time curve for gadolinium diethylenetriamine penta-acetic acid (Gd-DTPA) for each pixel (bottom) was derived from the changesin signal intensity over time (top). This curve is subsequently deconvolved on a pixel-by-pixel basis with the arterial input function AIF(t), calculated froma pixel placed within the middle cerebral artery (arrowed), and corrected for recirculation in order to calculate the maps of relative cerebral blood flow(curve height) and relative cerebral blood volume (area-under-the-curve).

844 T.J. O’Brien et al. / Journal of Clinical Neuroscience 14 (2007) 841–849

by a factor of three, and then the borders of the hippocam-pus manually outlined for each contiguous slice using amouse-driven cursor. Rigid anatomic landmarks were usedto define the hippocampal boundaries, that is posteriorlythe greatest length of fornix; medially the hippocampal fis-sure, then the uncal fissure anterior to the intralimbicgyrus, then the most mesial point of the ambient gyrusanterior to the uncal fissure, and anteriorly, as far as thehippocampus was reliably identified (generally down to lessthan 40 mm2). The hippocampal volume ratio (VR) wascalculated for the patients by dividing the side ipsilateralto the EEG lateralization by the contralateral side, andfor the controls by dividing the side with the smaller vol-ume by that with the larger volume.

2.7. T2 relaxometry quantification

The T2 relaxometry images were analyzed by placing acircular ROI at the center of the hippocampus on bothsides of each of the four mid-hippocampal slices. Carewas taken not to include any CSF or blood vessels thatlie very close to the hippocampus. The average ROI con-sisted of about 30 pixels. The T2 relaxometry ratios werecalculated for the patients by dividing the side ipsilateralto the EEG lateralization by the contralateral side, andfor the controls by dividing the side with the larger valueby that with the smaller value.

2.8. Statistical methods

Differences between the hippocampal ADC, CBF andHV ratios between the patients and the controls were testedusing the Mann-Whitney U-test (two-tailed). Correlationsbetween the ADC and CBF ratios and the HV and theT2 relaxometry ratios in the patients were tested usingSpearman’s R correlation. The level for determining statis-tical significance was set at p < 0.05 for all tests.

3. Results

3.1. Visual MRI assessment

The blinded reviewer, on visual assessment of the rou-tine images, detected the typical changes of unilateralMTS (that is, hippocampal atrophy and increased T2-sig-nal) in seven of the 10 patients and none of the control sub-jects. The visually detected hippocampal changes wereipsilateral to the ictal EEG lateralization in all cases.

3.2. Diffusion weighted image analysis

The ADC ratios for the patients (ipsilateral/contralat-eral) and controls (larger/smaller) are plotted in Fig. 3a.The ratios were significantly higher for the patients thancontrol subjects (median 1.11, range 1.0–1.35 vs. median

T.J. O’Brien et al. / Journal of Clinical Neuroscience 14 (2007) 841–849 845

1.03, range 1.01–1.07, p = 0.01, Mann-Whitney U-test). Sixof the 10 patients had ADC ratios that were outside therange of the control subjects. In all 10 patients the largerADC value was ipsilateral to the side of the ictal EEG on-set. The ADC ratios were outside the control range in oneof the three patients with normal visual MRI assessment.

3.3. Perfusion weighted image analysis

The side-to-side hippocampal rCBF ratios for the pa-tients (ipsilateral/contralateral) and controls (smaller/lar-ger) are plotted in Fig. 3b. The ratios were significantlylower for the patients than controls (median 0.61, range0.23–0.83 vs. median 0.88, range 0.84–0.96, p = 0.0005,Mann-Whitney U-test). All eight patients had rCBF ratiosthat were outside the range for the control subjects, and inall cases the smaller rCBF value was ipsilateral to the sideof the ictal EEG onset. This included two patients withnormal visual MRI assessments. The median side-to-side

Fig. 3. (a) The apparent diffusion co-efficient (ADC) ratios for 10temporal lobe epilepsy (TLE) patients (ipsilateral/contralateral) versus 10controls subjects (larger/smaller). The median values for the groups weresignificantly different, with six patients having ratios outside of the rangefor controls. The ratios for the patients were all greater than 1.0, indicatingconcordance with the EEG lateralization in all cases. (b) The regionacerebral blood flow (rCBF) ratios for eight TLE patients (ipsilateralcontralateral) versus nine control subjects (smaller/larger). The medianvalues for the groups were significantly different, with all of the patienthaving ratios outside of the range for controls. The ratios for the patientwere all < 1.0, indicating concordance with the EEG lateralization in alcases.

l/

ssl

(smaller/larger) ratios for the hippocampal rCBV measure-ments was also less for the patients than controls (median0.71, range 0.51–0.90 vs. median 0.87, range 0.77–0.99);however, two patients had rCBV ratios that fell withinthe control range, one of which was falsely lateralized(higher value ipsilateral to the EEG lateralization).

3.4. Correlation with quantitative hippocampal volume and

T2 relaxometry ratios

No significant correlations were found between the HVratios and either the ADC ratios (R = �0.03, p = 0.93,Spearman’s R) (Fig. 4a) or the rCBF ratios (R = 0.36,p = 0.39) (Fig. 4b). Similarly, no significant correlationwas found between the quantitative T2 mapping ratiosand the ADC ratios (R = 0.58, p = 0.08) (Fig. 4c) nor therCBF ratios (R = �0.40, p = 0.32) (Fig. 4d). Additionally,there was no significant correlation between the ADC andthe rCBF ratios (R = 0.19, p = 0.65).

4. Discussion

The results of this study demonstrate that both CEPIand DWI show promise as accurate and reliable tools forthe lateralization of ‘non-lesional’ TLE. The CEPI resultswere particularly encouraging, with all eight patients hav-ing a side-to-side rCBF ratio outside the range of the con-trol subjects and the side with the lowest value beingconcordant with seizure lateralization (Fig. 3b). Animalvalidation studies of CEPI have found a strong correlationbetween the measured rCBF and other measures including14C-iodoantipyrine perfusion imaging,37 and radioactivemicrospheres.38

In a previous small study of nine patients, the rCBV wasfound in all cases to be lower ipsilateral to the side of hip-pocampal atrophy and/or lateralization with [18F]fluorode-oxyglucose (FDG) PET.27 However, in our study we foundrCBV to be a less reliable lateralizing measure than rCBF(which was not evaluated in this previous study), withtwo patients having rCBV ratios within the control range,one of which was lower on the side contralateral to the epi-leptogenic side. Additionally, the previous study did notcompare the MRI perfusion data to the more standardquantitative MR measures: HV or T2-relaxometry.

Two preliminary studies have investigated endogenousMR perfusion methods that are based on arterial spinlabeling.39,40 In the first report the authors studied eight‘non-lesional’ TLE patients, and found that the side-to-sideMR perfusion ratios correlated well with the temporalblood flow asymmetries on H2

15O PET. However, no com-parison with control subjects was provided, nor was data inconcordance with the EEG lateralization. In the secondreport, 12 TLE patients and 12 control subjects were stud-ied and the MR perfusion side-to-side asymmetry indexcorrectly lateralized 11 of the TLE patients, but therewas considerable overlap with the range for the controlsubjects.40 The endogenous MR perfusion techniques have

Fig. 4. (a) The hippocampal volume (HV) ratios versus the apparent diffusion co-efficient (ADC) ratios for 10 temporal lobe epilepsy (TLE) patients. (b)The HV ratio versus the regional cerebral blood flow (rCBF) ratios (smaller/larger) for eight TLE patients. (c) The quantitative T2 relaxometry ratiosversus the ADC ratios. (d) The quantitative T2 relaxometry ratios versus the rCBF ratios (all ratios ipsilateral/contralateral).

846 T.J. O’Brien et al. / Journal of Clinical Neuroscience 14 (2007) 841–849

the potential advantage over CEPI of being cheaper andless invasive; however, at the cost of significantly worse sig-nal-to-noise ratio and spatial resolution that may limit theability to detect small changes.39

None of the previous studies of MR perfusion imagingin TLE have investigated whether these provide lateralizinginformation that is additional to that of established quan-titative MRI techniques. The finding in our study that thehippocampal rCBF ratios correctly lateralized two patientswith normal visual MRI assessments (including one whohad symmetrical hippocampal volume measurements) isencouraging. This suggests that CEPI may be able toprovide additional, rather than just confirmatory, laterali-zation to that of standard MRI techniques. This hope isfurther supported by the finding that there was no signifi-cant correlation between the hippocampal rCBF and thehippocampal volume ratios (Fig. 4b) nor the quantitativeT2 relaxometry ratios (Fig. 4d); however further studies

involving larger numbers of patients are required to deter-mine if the blood flow changes are truly independent of de-gree of volume loss and T2 signal increase.

The results of the analysis of the DWI images, althoughsomewhat less impressive than the CEPI results, still holdsignificant promise for this technique in the evaluation ofmedical refractory TLE. The side-to-side hippocampalADC ratios were outside the range of the control subjectsin six of the 10 patients studied, and in all 10 cases the sidewith the higher ADC was ipsilateral to the EEG seizure lat-eralization (Fig. 3a). One patient with a normal visual MRIassessment had hippocampal ADC ratios that were outsidethe control range, but not the two patients with symmetri-cal hippocampal volume measurements. The lack of corre-lation of the hippocampal ADC ratios with thehippocampal volumes (Fig. 4a) and with the T2 relaxome-try ratios (Fig. 4c) raises the prospect that the diffusionchanges are also independent of volume loss and T2 signal

Fig. 4 (continued)

T.J. O’Brien et al. / Journal of Clinical Neuroscience 14 (2007) 841–849 847

changes. However, again further studies involving largernumbers of patients are required to establish this.

Animal studies of focal status epilepticus in rats haveshown ADC to be decreased immediately post-ictally,and increased in the interictal period.11,12 Several othershave also examined the ability of ADC measures to lateral-ize patients with TLE, finding similar results to our study,with increased values on the epileptogenic side in patientswith hippocampal sclerosis.13–19 One study investigatedADC values in white matter of 15 patients with TLE,and observed that the changes in diffusivity were notrestricted to the temporal lobe but extend into other brainregions beyond the presumed the origin of seizure.17 Inaddition, isolated case reports of DWI being performedperi-ictally in human focal epilepsy have confirmed thatthere are transient focal decreases in the ADC measure-ments in some patients.41–43

The explanation for the changes in ADC demonstratedin TLE patients is currently uncertain. The rate of diffusionof water molecules is much less in living tissues than in free

water, due to physical restrictions from membranes, fibers,macromolecules (such as proteins) and dissolved organicmolecules which alter the viscosity of the water. Focalchanges in these barriers to water diffusion would resultin focal changes in the ADC. This effect has been mostextensively studied in stroke, where decreases in ADC areseen within minutes of the development of acute ische-mia.9,10 It is thought that this decrease in ADC results froma disruption of energy metabolism with a failure of ionpumps, with consequent influx of intracellular sodiumand water into the intracellular space where there are morebarriers to free water diffusion.9 In addition, more recentevidence suggests that the decreased ADC in acute ische-mia may also be due in part to decreases in cell membranepermeability, caused by collapse of transmembrane iongradients,44 as well as possibly by a decrease in energy-dependent cytoplasmic circulation or an increase in theintracellular water viscosity.45 However, in contrast, theADC is increased in chronically infarcted brain, whichmay reflect the effect of neuronal loss resulting in a

848 T.J. O’Brien et al. / Journal of Clinical Neuroscience 14 (2007) 841–849

proportionally larger extracellular space compared withintracellular space and allowing water to diffuse morefreely through the interstital space.46

The increased ADC interictally in TLE may at leastpartly reflect a loss of hippocampal cells. The loss of neu-rons and their axons disrupts the systematic structuralorganization in the hippocampus, and this may allow thewater molecules in the extracellular space to diffuse morerapidly, randomly causing an increase in the ADC. It ispossible that the seizures themselves have a functional ef-fect on the hippocampal cell physiology, which results inan alteration in water diffusion.

One possible non-physiological explanation for the ob-served increase in ADC is the inadvertent inclusion of somecerebrospinal fluid from the lateral ventricle on the sidewith hippocampal atrophy. However, we believe this is un-likely, due to the careful manual drawing of the ROI on theoblique coronal slices. In addition, the lack of correlationwith the degree of hippocampal volume loss and T2 relax-ometry measures, also argues against this explanation.

Finally, it should be acknowledged that while this studydemonstrated that CEPI and DWI show strong promise asuseful tools for the lateralization of TLE, the specificity ofthese methods for the intrahemispheric seizure localizationrequires further validation. It is possible that the changesmeasured by these methods result from the involvementof the hippocampus by seizures rather from the primaryepileptogenic pathology itself. If this is the case, extrahip-pocampal seizure onsets that secondarily spread to the hip-pocampus, may show similar changes. Future studies withlarger numbers of patients, particularly comparing thosewith extratemporal or neocortical TLE with those with me-sial TLE, should be able to better clarify this importantissue.

In conclusion, the results of this study demonstrated thatCEPI and DWI provide accurate lateralizing information inmedically refractory TLE. The ability to obtain these inde-pendent functional measures during the same examinationas the routine structural MRI acquisition, has significantpractical advantages over the traditional approach requir-ing the utilization of several different imaging modalities,such as PET and SPECT, in addition to the MRI.

Acknowledgement

This study was supported by an unrestricted grant fromParke-Davis (Australia) Pharmaceuticals. The authors aregrateful for the technical and software assistance of DrQing Yang, PhD, Physicist, Department of Radiology,University of Melbourne, Royal Melbourne Hospital, Vic-toria, Australia.

References

1. Engel Jr J, Van Ness P, Rasmussen T, et al. Outcome with respect toepileptic seizures. In: Engel J, editor. Surgical Treatment of the

Epilepsies. New York: Raven Press; 1993. p. 609–21.

2. Jack Jr CR, Sharbrough FW, Cascino GD, et al. Magnetic resonanceimage-based hippocampal volumetry: correlation with outcome aftertemporal lobectomy. Ann Neurol 1992;31:138–46.

3. Kuzniecky R, Burgard S, Faught E, et al. Predictive value ofmagnetic resonance imaging in temporal lobe epilepsy surgery. Arch

Neurol 1993;50:63–9.4. Berkovic SF, McIntosh AM, Kalnins RM, et al. Preoperative MRI

predicts outcome of temporal lobectomy: an actuarial analysis.Neurology 1995;45:1358–63.

5. Jack CR. Magnetic resonance imaging: Neuroimaging and anatomy.Neuroimaging Clin N Am 1995;5:597–622.

6. Cook MJ, Fish DR, Shorvon SD, et al. Hippocampal volumetric andmorphometric studies in frontal and temoral lobe epilepsy. Brain

1992;115:1001–15.7. Jackson GD. New techniques in magnetic resonance and epilepsy.

Epilepsia 1994;35:S2–S13.8. Williamson PD, Van Ness P, Weizer H, et al. Surgically

remediable extratemporal syndromes. In: Engel JJ, editor. Surgical

Treatment of the Epilepsies. 2nd ed. New York: Raven Press;1993. p. 65–76.

9. Warach S, Gaa J, Siewert B, et al. Acute human stroke studied bywhole brain echo planar diffusion weighted MRI. Ann Neurol

1995;37:231–41.10. Moseley ME, Kucharczyk J, Mintorovitch J, et al. Diffusion-

weighted MR imaging of acute stroke: correlation with T2-weightedand magnetic susceptibility-enhanced MR imaging in cats. Am J

Neuroradiol 1990;11:423–9.11. Zhong J, Petroff OAC, Prichard JW, et al. Changes in water diffusion

and relaxation properties of rat cerebrum during status epilepticus.Magn Reson Med 1993;30:241–6.

12. Righini A, Pierpaoli C, Alger JR, et al. Brain parenchyma apparentdiffusion coefficient alterations associated with experimental complexpartial status epilepticus. Magn Reson Imaging 1994;12:865–71.

13. Wieshmann UC, Clark CA, Symms MR, et al. Water diffusion in thehuman hippocampus in epilepsy. Magn Reson Imaging 1999;17:29–36.

14. Hugg JW, Butterworth EJ, Kuzniecky RI. Diffusion mapping appliedto mesial temporal lobe epilepsy. Preliminary observations. Neurology

1999;53:173–6.15. Rugg-Gunn FJ, Eriksson SH, Symms MR, et al. Diffusion tensor

imaging of cryptogenic and acquired partial epilepsies. Brain

2001;124:627–36.16. Assaf B, Mohamed F, Abou-Khaled K, et al. Diffusion tensor

imaging of the hippocampal formation in temporal lobe epilepsy.AJNR Am J Neuroradiol 2003;24:1857–62.

17. Arfanakis K, Hermann B, Rogers B, et al. Diffusion tensor MRI intemporal lobe epilepsy. Magn Reson Imaging 2002;20:511–9.

18. Eriksson S, Rugg-Gunn F, Symms M, et al. Diffusion tensor imagingin patients with epilepsy and malformation of cortical development.Brain 2001;124:617–26.

19. Thivard L, Lehericy S, Krainik A, et al. Diffusion tensor imaging inmedial temporal lobe epilepsy with hippocampal sclerosis. Neuroim-

age 2005;28:682–90.20. Edelman R, Mattle HP, Atkinson DJ, et al. Cerebral blood flow:

Assessment with dynamic contrst-enhanced T2*-weighted MR imag-ing at 1.5T. Radiology 1990;176:211–20.

21. Belliveau J, Rosen BR, Rzedzian RR, et al. Functional cerebralimaging by susceptibility-contrast NMR. Magn Reson Med

1990;14:538–46.22. Berkovic SF, Newton M, Rowe C. Localization of epileptic foci using

SPECT. In: Luders H, editor. Epilepsy Surgery. New York: RavenPress; 1992. p. 251–6.

23. Newton MR, Berkovic S. Interictal, ictal, and postictal single-photonemission computed tomography. In: Cascino G, Jack C, editors.Neuroimaging in Epilepsy: Principles and Practice. Boston: Butter-worth-Heinemann; 1997. p. 177–92.

24. Jack Jr CR, Mullan BP, Sharbrough FW, et al. Intractable nonle-sional epilepsy of temporal lobe origin: Lateralization by interictalSPECT verses MRI. Neurology 1994;44:829–36.

T.J. O’Brien et al. / Journal of Clinical Neuroscience 14 (2007) 841–849 849

25. Spencer SS. The relative contributions of MRI, SPECT, and PETimaging in epilepsy. Epilepsia 1994;35 (Suppl 6):S72–89.

26. Aronen HJ, Gazit IE, Louis DN, et al. Cerebral blood volume mapsof gliomas: comparison with tumor grade and histological findings.Radiology 1994;191:41–51.

27. Wu RH, Bruening R, Noachtar S, et al. MR measurements ofregional relative cerebral blood flow in epilepsy. J Mag Reson Imaging

1999;9:435–40.28. Warach S, Levin J, Schomer D, et al. Hyperperfusion of ictal seizure

focus demonstrated by MR perfusion imaging. AJNR Am J Neuro-

radiol 1994;15:965–8.29. Stejskal EO, Tanner JE. Spin diffusion measurements: spin-echoes in

the presence of a time-dependent field gradient. J Chem Physiol

1965;42:288–92.30. Ulug AM, Beauchamp Jr N, Bryan RN, et al. Absolute quantitation

of diffusion contrasts in human stroke. J Chem Physiol 1997;42:288–92.

31. Van Gelderen P, De Vleeschouwer MHM, DesPres D, et al.Water diffusion and acute stroke. Magn Reson Med 1994;31:154–63.

32. Weisskoff RM, Zuo CS, Boxerman JL, et al. Microscopic suscepti-bility variation and transverse relaxation: theory and experiment.Magn Reson Med 1994;31:601–10.

33. Villringer A, Rosen BR, Belliveau J, et al. Dynamic imaging withlanthanide chelates in normal brain: contrast due to magneticsusceptibility effects. Magn Res Med 1988;6:164–74.

34. Ostergaard L, Weisskoff RM, Chesler DA, et al. High resolutionmeasurement of cerebral blood flow using intravascular tracer boluspassages. Part I: a mathematical approach and statistical analysis.Magn Reson Med 1996;36:715–25.

35. Ostergaard L, Sorensen AG, Kwong KK, et al. High resolutionmeasurement of cerebral blood flow using intravascular tracer boluspassages. Part II: experimental comparison and preliminary results.Magn Reson Med 1996;36:726–36.

36. Cook MJ. Mesial temporal sclerosis and volumetric investigations.Acta Neurol Scand 1994;152:109–14.

37. Wittlich F, Kohno K, Mies G, et al. Quantitative measurement ofregional blood flow with gadolinium diethyenetriaminepentaacetatebolus track NMR imaging in cerebral infarcts in rats: validation withthe iodo [14C]antipyrine technique. Proc Natl Acad Sci USA

1995;92:1846–50.38. Muller TB, Jones RA, Haraldseth O, et al. Comparison of MR

perfusion imaging and microsphere measurements of regional cerebralblood flow in a rat model of middle cerebral artery occlusion. Magn

Reson Imaging 1996;14:1177–83.39. Liu H-L, Kochunov P, Hou J, et al. Perfusion-weighted imaging of

interictal hypoperfusion in temporal lobe epilepsy using FAIR-HASTE: Comparison with H2

15O PET measurements. Magn Reson

Med 2001;45:431–5.40. Wolf RL, Alsop DC, Levy-Reis I, et al. Detection of mesial temporal

lobe hypoperfusion in patients with temporal lobe epilepsy by use ofarterial spin labeled perfusion MR imaging. AJNR AM J Neuroradiol

2001;22:1334–41.41. Diehl B, Najm I, Ruggieri P, et al. Periictal diffusion-weighted

imaging in a case of lesional epilepsy. Epilepsia 1999;40:1667–71.42. Wieshmann UC, Symms MR, Shorvon SD. Diffusion changes in

status epilepticus. Lancet 1997;350:493–4.43. Diehl B, Najm I, Ruggieri P, et al. Postictal diffusion-weighted

imaging for the localisation of focal epileptic areas in temporal lobeepilepsy. Epilepsia 2001;42:21–8.

44. Helpern JA, Huang N. Diffusion-weighted imaging in epilepsy. Magn

Reson Imaging 1995;13:1227–31.45. Duong TQ, Ackerman JJ, Ying HS, et al. Evaluation of extra- and

intracellular apparent diffusion in normal and globally ischemic ratbrain via 19F NMR. Magn Reson Med 1998;40:1–13.

46. Knight RA, Dereski MO, Helpern JA, et al. Magnetic resonanceimaging assessment of evolving focal cerebral ischaemia. Stroke

1994;25:1252–62.