Predicting
success
of
vagus
nerve
stimulation
(VNS)
from
interictal
EEG
C.C.
de
Vos
a,b,*
,
L.
Melching
a,
J.
van
Schoonhoven
a,
J.J.
Ardesch
b,
A.W.
de
Weerd
c,
H.C.E.
van
Lambalgen
c,
M.J.A.M.
van
Putten
a,ba
MIRAInstituteforBiomedicalTechnologyandTechnicalMedicine,UniversityofTwente,TheNetherlands
b
DepartmentofClinicalNeurophysiology,MedischSpectrumTwente,TheNetherlands
c
SEINEpilepsieInstellingen,Zwolle,TheNetherlands
1. Introduction
Vagusnerve stimulation(VNS) hasshown tobean effective treatment for epilepsy in numerous patients. Most long-term studiesthatweredonetoassesstheefficacyofVNSconcludedthat amorethan50%seizurereductionwasaccomplishedin20–55%of the patients after treatment for six months to six years.1–3
According to Janszky et al.,4 0–24% ofthe medically refractory
patientstreatedwithVNSbecomesseizurefree.
DespitethegrowingapplicationofVNS,itisstillnotpossibleto predictwhichpatients respond towhatextenttoVNStherapy. DeterminingthesuccessofVNSisimportanttocounselpatients andgivetheminformationabouttheexpectedseizurereduction. Potentialrespondersmightnotneedtotryotherkindsoftherapy beforetheyreceiveaneffectiveVNSsystemandontheotherhand, alowlikelihoodtorespondcouldpreventsomeonefromhavingan expensiveVNSsystemimplantedwhileonlyminimaleffectswill beobtained.
Moststudies thatattempttopredictthesuccessof VNSare based upon the localization of the seizure focus, patient characteristics or epilepsy syndrome. However, predictors of success are still elusive. Itwas foundthat VNSresponsiveness wasassociated witholderageandlonger epilepsyduration5 or
rathertobeindependentofepilepsyduration6andassociatedwith younger age.7 VNS successwas foundto berelatedtoepilepsy
syndromes other than Lennox-Gastaut syndrome5 or rather to
Lennox-Gastaut syndrome3,8 and tonic seizures.3 Furthermore, Scherrmannetal.9concludedthatseizureoutcomewaspositively
correlated with VNS duration and Handfort et al.10 found that
seizurereductionwaspositivelycorrelatedwithhighstimulation settings.
Untilnow,veryfewstudiesevaluatedwhethersuccessofVNS canbeforecastedusingtheelectroencephalogram(EEG).4,11There
are somemoreepilepsysurgery studiesusing EEGasa tool to assess outcome prognosis.12–14 These studies are all primarily
basedonthevisualanalysisoftheEEG,forinstancebycountingthe numberofInterictalEpilepticDischarges(IED)beforeonsetofthe therapy.Janszkyetal.4showedthatabsenceofbilateralIEDsinthe
EEGbeforeVNSimplantationwasassociatedwithaseizurefree outcome.
However, observing the different wave-formsin the EEG is subjective and laborious because the results depend on the A R T I C L E I N F O
Articlehistory:
Received20October2010
Receivedinrevisedform3March2011 Accepted1April2011
Keywords: Epilepsy
Quantitativeelectroencephalography (qEEG)
Brainsymmetryindex(BSI) Vagusnervestimulation(VNS)
A B S T R A C T
Purpose:Vagus nervestimulation(VNS)hasshownto be aneffective treatmentfordrugresistant epilepsyinnumerouspatients,however,notinall.Itisstillnotpossibletopredictwhichpatientswill profitfromVNS.Inthispilotstudy,weexplorepredictiveinterictalEEGfeaturesforseizurereduction afterVNS.
Methods:19PatientswithmedicallyrefractoryepilepsyandanimplantedVNSsystemwereincluded. InterictalEEGregistrations,recordedbeforeimplantation,wereretrospectivelyanalysed.Aquantative symmetrymeasure,thepairwisederivedbrainsymmetryindex(pdBSI),wastestedtopredictVNS outcome.Reductioninseizurefrequencywasusedtodefinetheresponders.
Results:10PatientsdidrespondtoVNS,ofwhom7patientshadaseizurereductionofatleast50%ina follow-upperiodof2years.Onaverage,wefindhigherpdBSIvaluesfordelta,theta,alphaandbetabands for non-responders thanfor responders. The average pdBSI of the theta and alpha bands could significantlydiscriminatebetweenrespondersandnon-responders.
Conclusion:In this study, quantifying EEG symmetryusing the pdBSI shows promisingresults in predictingthereductionofseizurefrequencyafterVNStreatment.
ß2011BritishEpilepsyAssociation.PublishedbyElsevierLtd.Allrightsreserved.
*Correspondingauthorat:MedischSpectrumTwente,DepartmentofClinical Neurophysiology,P.O.Box50,000,7500KAEnschede,TheNetherlands. Tel.:+31534873532;fax:+31534872816.
E-mailaddress:c.devos@mst.nl(C.C.deVos).
ContentslistsavailableatScienceDirect
Seizure
j o urn a lhom e pa g e :ww w . e l se v i e r. c om / l oca t e / y se i z
1059-1311/$–seefrontmatterß2011BritishEpilepsyAssociation.PublishedbyElsevierLtd.Allrightsreserved. doi:10.1016/j.seizure.2011.04.002
individualneurophysiologists’ experienceandexpertise. Quanti-tative EEG (qEEG) analysis may partially replace the visual interpretation15–17andisamoreobjectiveandperhapssensitive
methodthanvisualinterpretationofEEG.qEEGmayevendetect characteristicsoftheEEGthatarenotvisibleforthenakedeye,like synchronization measures, power per frequency band and symmetrymeasures.
Wehypothesizethatsymmetrycanbearelevantfeatureto predict the effect of VNS therapy. Van Putten18,19 originally
proposed the brain symmetry index (BSI) as a measure for electroencephalographicsymmetry.TheBSIquantifiesthespatial EEG symmetry and has found clinical applications for the detection of (focal) ischemia20,21 and focal seizure activity.22
Wehypothesizethattheinterictal EEGfrompatientssuffering from(multi-focal)pharmacoresistantepilepsymaybe character-ized by an increased asymmetry. This is motivated by the observationthat in manyof these patients, the interictal EEG oftenshowsasymmetricfeatures,e.g.focalslowingoramplitude asymmetries. In this study, therefore, we explore whether baselineEEGsymmetry,asquantifiedbytheBSI,isapredictor forsuccessofVNStherapy.
2. Materialsandmethods 2.1. Patientselection
VNStreatedpatientswereselectedretrospectively(seeTable 1).Allpatientssufferedfrom(multi)focal,medicallyintractable epilepsy with varying focus locations and were scheduled for implantationof a vagusnerve stimulator(Cyberonics, Houston, TX) between 2001 and 2008. All patients were treated at the epilepsycentreSEINZwolle.Patientswereaged16yearsorolder andshouldhavehadanEEGrecordedshortlybeforetheonsetof VNS therapy. During this EEG recording, no epileptic seizure shouldhaveoccurred,astheinterictalEEGpatternisanalysedand sufficientminutesofartefactfreeEEGshouldbeavailable.Three months prior to implantation and during the first year after implantation anticonvulsant drug intake should have been unchanged.
Patientsandtheirfamilyshouldhavekeptseizurediariesfor oversixmonthspriortotheVNStherapyandduringthefirstoneto twoyearsofVNStherapy.Theevaluatedseizurereductionduring therapywasusedtodeterminethesuccessofVNS.Theaverage numberofseizurespermonthwascalculatedandtwodefinitions were used to define responders. Responders0 were defined as
havinganyreductioninseizurefrequencyandresponders50hadat
least50%seizurereduction.
Thestimulationparametersarepersonalizedforeachpatient. Oftenthetherapystartedwithastimulationcycleof30sonand 5minoff.Theamplitudewasincreasedguidedbytheeffectsand side-effects of the stimulation. When further increase of the amplitude was neither effective nor possible, a more rapid stimulationcyclewastested.
2.2. EEGanalysis
All EEGs were recorded by trained personnel at the SEIN epilepsycentre. Electrodes wereplaced according tothe10/20 system,fixedbyadhesiveandconductivegel.Impedancesdidnot exceed10k
V
.Samplingfrequencywas200Hz.Twentytoforty minutes ofEEG wererecorded,according tostandard protocol, containing periods of hyperventilation, eyes closed, eyes open, intermittentphotostimulationandsomatosensorystimulationof thehand.Source referencewasused for montage and onlyperiods of closedeyeswithoutanyformofstimulationwereusedforanalysis. ThiswasdonetoavoidqEEGabnormalitiesduetoeyemovements orotherprovocations.Furthermore,periodswithIEDs,movement artefacts or periods indicating drowsiness were excluded after visual inspection.At least several minutes of artefact free EEG activitywithclosedeyesneededtobepresentforanEEGrecording tobeanalysed.Selectedepochsweresubsequentlyfilteredwitha bandpassfilterbetween0.5and30Hz.Epochsof400sampleswith 50% overlapwereFouriertransformedwithpwelchin MATLAB (TheMathWorks,Inc.)usingaHammingwindow.
2.3. Features
AnewimplementationoftheBSIwasusedforanalysingbrain symmetry.OriginallytheBSIwasproposedasameasureforthe mean electroencephalographicspatial symmetryofthe brain.19
Morerecently,thepdBSIwasintroducedbySheorajpandayetal.21
asanaturalextensionoftheBSI.ThepdBSIisasinglechannelpair wisederivedBSIwhichevaluatesasymmetryalonghomologous channel pairs instead of globalasymmetry, which is measured withtheBSI. Comparisonof homologouschannel pairs (pdBSI) insteadofglobalhemisphericdifferences(BSI)couldleadtoamore sensitive determination of abnormal asymmetry in epilepsy patientswithseveralorcryptogenicfoci.ThepdBSI21iscalculated with:
Table1
Patientcharacteristicsatonsetofvagusnervestimulation.
No. Sex Age(y) Typeepilepsy EffectVNS(%reduction) pdBSIdbaseline pdBSItbaseline pdBSIabaseline
1 M 30 Focal 0 0.330 0.304 0.292 2 M 47 Multifocal 50 0.320 0.288 0.303 3 F 56 Focal 60 0.320 0.295 0.281 4 M 16 Multifocal 0 0.312 0.306 0.300 5 M 21 Multifocal 0 0.339 0.327 0.320 6 M 50 Focal 25 0.289 0.287 0.304 7 F 55 Focal 80 0.308 0.286 0.292 8 F 46 Focal 0 0.441 0.435 0.395 9 F 33 Focal 30 0.370 0.342 0.370 10 F 31 Multifocal 0 0.398 0.364 0.325 11 M 41 Focal 60 0.299 0.267 0.271 12 F 63 Focal 0 0.273 0.233 0.243 13 M 42 Focal 0 0.332 0.354 0.326 14 F 39 Focal 50 0.278 0.272 0.281 15 F 47 Focal 50 0.287 0.269 0.259 16 M 64 Focal 80 0.322 0.296 0.302 17 M 45 Focal 25 0.322 0.281 0.335 18 M 29 Focal 0 0.340 0.308 0.306 19 M 16 Focal 0 0.311 0.307 0.315
pdBSI¼1 N XK i¼1 1 M XM j¼1 Ri;jLi;j Ri;jþLi;j
WithRi,jandLi,jtheFouriercoefficientbelongingtofrequency
i=1,...,Kofrightandlefthemisphericbipolarderivationsj=1,2, ...,MforNdiscretetimepoints.Thebipolarderivationsusedto calculatethepdBSIareFp1–Fp2,F3–F4,F7–F8,F9–F10,C3–C4,P3– P4,P7–P8,O1–O2,T7–T8.
TheBSIisdefinedin sucha waythatthelowertheBSI,the highertheEEGsymmetry.EEGrecordingsfrom‘‘normalhealthy brains’’resultin BSIvalues ofabout0.0719andpdBSIvaluesof
about0.13.21IncreasedpdBSIvaluesareobtainedifEEGrecordings
showasymmetries,e.g.focalslowingordifferencesinamplitudes ofhomologouschannelpairs.
Four different frequency ranges were explored: delta (0.5– 4Hz),theta(4–8Hz),alpha(8–12Hz)andbeta(12–30Hz).Higher frequencyrangeswerenotinvestigatedbecauseofthelowsignal tonoiseratioforthesefrequencieswhenusingscalpEEG. 2.4. Statistics
For thediscriminationbetween responders and non-respon-dersbasedonsinglefeatures,t-testswereappliedwhendatawas normallydistributed;Mann–WhitneyUtestswereusedwhendata wasnotnormallydistributed. Normalitytestsweredoneforall singlefeaturesusingtheShapiroWilktest.Aconfidenceintervalof 95%wasusedtodefinesignificantresults.Thiscurrentstudywas usedtodiscoverpromisingfeaturesthatwillbetestedextensively inourfuturestudy.
3. Results
TheEEGrecordingsofonly19patientsmatchedourinclusion criteriaandwereanalysed.TheseEEGswererecordedwithinan interval of six weeks prior to onset of the VNS therapy and containedonlyfewartefacts.Thepatientsincludedwere11men and8women,withameanageof45years(standarddeviation15 years).15patientssufferedfromfocalepilepsy,while4hadmulti focal epilepsy. Due to VNS therapy, 10 patients obtained a reduction in seizure frequency (responders0)of which 7 had a
reducedseizurefrequencyofatleast50%(responders50).Forall
patients,thepdBSIvaluesofallfrequencybandswereobviously increasedcomparedtotheaveragevalueof0.13forhealthycontrol subjectsmeasuredby Sheorajpandayet al.21(seeTable 1). This indicatesthatbeforetreatmentwithVNS,theasymmetryofthe EEGactivity for allthe patients is significantly higherthan for healthycontrols.
Table 2 shows the p-values for the discrimination between
responders0andnon-responders0andbetweenresponders50and
non-responders50 using the pdBSI for the different frequency
bands. A lower pdBSI value indicates more symmetry along homologouschannelpairsofthetwohemispheres.ThepdBSIfor thethetaband(pdBSIt)issignificantlylowerforresponders0than
for non-responders0 and thepdBSI forthe deltaband(pdBSId)
showsadifferencebetweenresponders0andnon-responders0as
well,howeverthisdifferenceislessconvincing.ThepdBSIforthe alphaband(pdBSIa)issignificantlylowerforresponders50thanfor
non-responders50andboththepdBSIdandpdBSItalsoshowclear
differencesbetweenresponders50andnon-responders50.
Fig.1showsthepdBSIvaluesforresponders50andresponders0
forthefourdifferentfrequencybands.Respondershaveonaverage lowerpdBSIvaluesforallfrequencyrangesthannon-responders, independent of thedefinitionof responders. Relatively lowBSI values for responders imply that, prior to the onset of VNS treatment,symmetryintheEEGisyethigherforrespondersthan fornon-responders.
Resultsshowthatresponders50haveslightlylowerBSIvalues
thanresponders0.Inaddition,thevarianceinthepdBSIofthe
non-respondersisingeneralmuchlargerthanthevarianceinthepdBSI of the responders. Differences between responders50 and
non-responders50 are larger than between responders0 and
non-responders0.
4. Discussion
Inthisstudy,weevaluateparticularsymmetrymeasuresofthe EEGintheirabilitytopredictwhetherpatientswillrespondwitha seizurereductiontovagusnervestimulation.Itisfoundthatlocal symmetryvalues(pdBSI),especiallyforthedelta,thetaandalpha frequency ranges, correlate with a positive response to VNS treatment.AlthoughinpatientsthepdBSIvaluesarehigherthanin healthycontrols,non-respondershavesignificantlyhigherpdBSI valuesbeforetheonsetofVNStreatmentthanresponders.
In many patients suffering fromepilepsy, in particular focal seizures,theinterictalEEGmayshowlocalabnormalities,e.g.focal Table2
p-Valuesfordiscriminationbetweenrespondersandnon-responders,basedonpair wisederivedbrainsymmetryvalues(pdBSI)fordifferentfrequencybandsextracted fromEEGsrecordedbeforetheonsetofVNStreatment.
pdBSI(delta,theta,alpha,beta) pdBSId pdBSIt pdBSIa pdBSIb Non-responders0vs.responders0 0.09 0.01* 0.31 0.94
Non-responders50vs.responders50 0.07 0.06 0.02* 0.14 *
p-Value<0.05.
Fig.1.Differencesbetween(a)responders50(n=7)andnon-responders50(n=12)
and(b)responders0(n=10)andnon-responders0(n=9)inpdBSIforthedelta,
theta,alphaandbetaband,extractedfromEEGsrecordedbeforeonsetofVNS treatment.Errorbarsrepresentthestandarddeviation.*p<0.1,**p<0.05.
slowingordifferencesinamplitude,whichwillincreasethepdBSI values.Inthisstudy,weindeedfoundthatforallpatientsthepdBSI values are higher than for healthy controls. In addition, non-respondersshowedsignificantlyhigherpdBSIvaluesthan respon-ders. Thissuggests that in patients withonly mildly abnormal EEGs,asreflectedbythemoderateincreasesinthepdBSI,VNSis morelikelytohaveaclinicallyrelevanteffect.Astheamountof asymmetry as quantifiedby the pdBSI, may indeed reflect the deviationfromnormal,VNSseemsrelativelymoreeffectiveinmild cases.
Itwasalsofoundthat,usingtheBSItoquantifyhemispheric symmetry,wecanbetterpredictwhether apatientwillhave a seizurereductionofmorethan50%(responder50),thanthatwecan
predictany seizure reduction at all (responder0). Patients who
have experienced a seizure reduction of less than 50% were patients with limited response to the therapy. Including these poor-responders in a statistical analysis decreases the power. Although our study population is too small to differentiate betweenmoregradualresponses,itappearsthatbrainsymmetry valuesaremostsuitedtoidentifythebestresponders.
Anticonvulsantdrugsmayinfluenceseveralcharacteristicsof theEEG.Thepatientsinthisstudyallusedcombinationsoftwoto fourdifferentanticonvulsantdrugs.Manypatientsused combina-tions including valproic acid, carbamazepine or lamotrigine. However,theeffectsofanticonvulsantsontheEEGaretypically minorandglobal ratherthan focal,e.g. a minimalreduction in alphameanfrequency.23Therefore,theeffectofanticonvulsants
ontheBSIwillbeminimal,ifpresentatall.Indeed,acorrelation betweenpdBSIvaluesandparticularanticonvulsantdrugswasnot found.
Evaluation of the interictal EEG to predictsuccess of vagus nerve simulation was previously briefly described by Majoie etal.11whomentionedaloweraveragebackgroundfrequencyand
moreinterictalepileptiformabnormalitiesinpatientswhodidnot respondtoVNStreatment.Theobtaineddifferenceinbackground frequencyintheirstudywashowevernotstatisticallysignificant. Research done by Janszky et al.4 showed that the absence of
bilateralIEDspredictedsuccessofVNStreatment.Quantification oftheEEGpatternsinbothpreviousstudieswasdifficult,asthe authorsusedprimarilyvisualinspectionof theEEGrecordings, whereas ourapproach is alsofeasible when there are no IEDs visibleandallowsmoreobjectivequantificationofpredictiveEEG characteristics.
Various related studies usedqEEG features to evaluateVNS therapy,ratherthantopredictitseffects.24–26Inthesestudiesof
theinterictal EEG,low EEG frequencies (i.e.delta–theta–alpha) havebeenfoundtobemostrelevant,andassociatedwithvarious epilepsysyndromes.Novaketal.26foundthatthechroniceffectof
VNSontheinterictalEEGshowedatrendtowardsdecreaseddelta powerandincreasedalphapower. Marrosuetal.25investigated
backgroundinterictalEEGinawakeepilepsypatientsaftertherapy aswell.Theyevaluatedbothchangesinthepowerspectrumand thesynchronizationlevelusingthecrossspectraldensityfunction normalized by individual auto-spectral density functions.27 In
elevensubjects,theintra-andinter-hemisphericsynchronization inthethetaband(4–7.5Hz)bothdecreased,whereasthepowerof thegammaband(20–50Hz)andtheintra-hemispheric synchro-nization increased. Since evaluation of qEEG changes is very relevantandmayprovideimportantinformationaboutpotential mechanismsofVNStherapy,wehavealsostudiedqEEGchanges dueto VNStherapy. However, this is beyond thescope of this paper.Wehavechosentofocusonprediction,aspredictingthe responsetoVNSwillprovidegreatclinicalbenefits.Determining features that will predict potential responders and provide patientswithinformationabouttheexpectedseizurereduction willincreasetheefficacyofVNStherapy.
ThisstudyisafirstexplorationofpredictiveqEEGfeaturesto identifyresponderstoVNStherapy.Althoughwewerenotableto predicttheeffectofVNSfortheindividualpatient,EEGsymmetry quantifiedbypdBSIseemsarelevantfeature.Additionalvalidation oftheproposedBSIfeaturesandthecreationofapredictionmodel aresubjectsofourfuturestudy.
5. Conclusion
Insum,ourstudyshowsthatEEGsymmetryfeaturesmaybe usefultopredictthesuccessofVNStherapy.Brainsymmetryindex valuesin thedifferentfrequencybands showthatpatients who will not respond to VNS treatment have, on average, more asymmetric spectral characteristics of the interictal EEG than responders.
Acknowledgments
WewouldliketothanktheDepartmentofClinical Neurophys-iologyatSEINZwollefortheregistrationoftheEEGs.
WeconfirmthatwehavereadtheJournal’spositiononissues involved in ethical publication and affirm that this report is consistentwiththoseguidelines.
Conflictofinterest
Noneoftheauthorshasanyconflictofinteresttodisclose. References
1.SchachterSC,SchmidtD.Vagus nerve stimulation.secondedition.MartinDunitz Publishers;2003.
2.ArdeschJJ,BuschmanHPJ,Wagener-SchimmelLJJC,VanderAaHE,HagemanG. Vagusnervestimulationformedicallyrefractoryepilepsy:alongterm follow-upstudy.Seizure2007;16:579–85.
3.ShahwanA,Bailey.MaxinerW,HarveyAS.Vagusnervestimulationfor refrac-toryepilepsy inchildren: moretoVNS thanseizurefrequencyreduction. Epilepsia2009;50:1220–8.
4.JanszkyJ,HoppeM,BehneF,TuxhornI,PannekHW,EbnerA.Vagusnerve stimulation: predictors of seizure freedom. J Neurol Neurosurg Psychiatry 2005;76:384–9.
5.LabarD.Vagusnervestimulationfor1yearin269patientsonunchanged antiepilepticdrugs.Seizure2004;13:392–8.
6.TecomaES,IraguiVJ.Vagusnervestimulationuseandeffectinepilepsy:what havewelearned?EpilepsyBehav2006;8:127–36.
7.GhaemiK,ElsharkawyAE,SchulzR,HoppeM,PolsterT,PannekH,etal.Vagus nervestimulation:outcomeandpredictorsofseizurefreedominlong-term follow-up.Seizure2010;19:264–8.
8.FrostM,GatesJ,HelmersSL,WhelessJW,LevisohnP,TardoC,et al. Vagusnerve stimulationinchildrenwithrefractoryseizuresassociatedwith Lennox-Gas-tautsyndrome.Epilepsia2001;42:1148–52.
9.ScherrmannJ,HoppeC,KralT.Vagusnervestimulation.Clinicalexperienceina largepatientserie.JClinNeurophysiol2001;18:408–14.
10.HandfortA,DeGiorgioCM,SchachterSC.Vagusnervestimulationtherapyfor partial-onset seizures. a randomized active-control trial. Neurology 1998;51:48–55.
11.MajoieHJM,BerfeloMW,AldenkampAP,RenierWO,KesselsAGH.Vagusnerve stimulationinpatientswithcatastrophicchildhoodepilepsy,a2-year follow-upstudy.Seizure2005;14:10–8.
12.JanszkyJ,JokeitH,SchulzR,HoppeM,EbnerA.EEGpredictssurgicaloutcomein lesionalfrontallobeepilepsy.Neurology2000;54:1470–6.
13.SchulzR,LudersHO,HoppeM,TuxhornI,MayT,EbnerA.InterictalEEGand ictalscalpEEGpropagationarehighlypredictiveofsurgicaloutcomeinmesial temporallobeepilepsy.Epilepsia2000;41:564–70.
14.LeeSA,YinSB,LimYM,KangJK,LeeJK.Factorspredictingseizureoutcomeof anterior temporallobectomy for patients with mesial temporalsclerosis. Seizure2006;15:397–404.
15.NuwerM.AssessmentofdigitalEEG,quantitativeEEG,andEEGbrainmapping. Neurology1997;49:277–92.
16.FinniganSP,WalshM,RoseSE,ChalkJB.QuantitativeEEGindicesofsub-acute ischaemic stroke correlate with clinical outcomes. Clin Neurophysiol 2007;118:2525–32.
17.vanPuttenMJAM.Thecolorfulbrain:visualizationofEEGbackgroundpatterns. JClinNeurophysiol2008;25:63–8.
18.vanPuttenMJAM.ExtendedBSIforcontinuousEEGmonitoringincarotid endarterectomy.ClinNeurophysiol2006;117:2661–6.
19.van Putten MJAM. The revised brain symmetry index. Clin Neurophysiol 2007;118:2362–7.
20.vanPuttenMJAM,Tavy DLJ.ContinuousquantitativeEEGmonitoringin hemisphericstrokepatientsusingthebrainsymmetryindex.Stroke2004; 11:2489–92.
21.SheorajpandayRVA,NagelsG,WeerenJTM,vanPuttenMJAM,deDeynPP. ReproducibilityandclinicalrelevanceofquantitativeEEGparametersin cere-bralischemia:abasicapproach.ClinNeurophysiol2009;120:845–55. 22.vanPutten MJAM,Kind T,VisserF, LagerburgV.Detectingtemporallobe
seizuresfromscalpEEGrecordings:acomparisonofvariousfeatures.Clin Neurophysiol2005;116:2480–9.
23.ClemensB,Me´nesA,PirosP,BessenyeiM,AltmannA,JerneyJ,et al. effectsof carbamazepine,oxcarbazepine,valproate,lamotrigine,andpossible clinical relevanceofthefindings.EpilepsyRes2006;70:190–9.
24.SalinskyMC,BurchielKJ.VagusnervestimulationhasnoeffectonawakeEEG rhythmsinhumans.Epilepsia1993;34:299–304.
25.MarrosuF,SantoniF,PulighedduM,BarberiniL,MaleciA,EnnasF,et al. Increase in20–50hz(gammafrequencies)powerspectrumandsynchronizationafter chronicvagalnervestimulation.ClinNeurophysiol2005;116:2026–36. 26.NovakK,HoeppnerJ,RistanovicRK,BernsteinLP,TaberJ,CozzensJ.Theeffects
ofvagusnervestimulation(VNS)therapyoninterictalepileptiformdischarges andneuropsychologicalperformance.Epilepsia2006;47:333.
27.NunezPL,SrinivasanR,WestdorpAF,WijesingheRS,TuckerDM,SilbersteinRB, et al. EEGcoherencyI:statistics,referenceelectrode,volumeconduction, laplacians,corticalimaging,andinterpretationatmultiplescales. Electroence-phalogrClinNeurophysiol1997;103:499–515.