Citation/Reference Van de Vel A., Cuppens K., Bonroy B., Milosevic M., Jansen K., Van Huffel S., Vanrumste B., Lagae L., Ceulemans B., ``Non-EEG seizuredetection systems and potential SUDEP prevention : State of the art.'', Seizure, vol.
22, 2013, pp. 345-355
Archived version Final publisher’s version / pdf
Published version insert link to the published version of your paper http://dx.doi.org/10.1016/j.seizure.2013.02.012
Journal homepage insert link to the journal homepage of your paper http://www.elsevier.com/locate/yseiz
IR https://lirias.kuleuven.be/handle/123456789/402545
(article begins on next page)
Review
Non-EEG seizure-detection systems and potential SUDEP prevention: State of the art
AnoukVan de Vela,*,KrisCuppensb,c,d,Bert Bonroyb, MilicaMilosevicc,d,KatrienJansene, SabineVan Huffelc,d,Bart Vanrumsteb,c,d,LievenLagaee,f, BertenCeulemansa,f
aDepartmentofNeurology-PediatricNeurology,AntwerpUniversityHospital-UniversityofAntwerp,Wilrijkstraat10,B-2650Edegem,Belgium
bMobilab,K.H.KempenUniversityCollege,Kleinhoefstraat4,B-2440Geel,Belgium
cDepartmentofElectricalEngineering-ESAT,SCD-SISTA,KULeuven,KasteelparkArenberg10postbus2440,B-3001Heverlee,Belgium
diMindsFutureHealthDepartment,KasteelparkArenberg10postbus2440,B-3001Heverlee,Belgium
eDepartmentofPediatricNeurology,UniversityHospitalsLeuven–CampusGasthuisberg,Herestraat49,B-3000Leuven,Belgium
fEpilepsyCentreforChildrenandYouthPulderbos,Reebergenlaan4,B-2242Zandhoven,Belgium
1. Introduction
Epilepsyisthemostcommonseriousneurologicalcondition, affecting almost 60 million people (close to 1% of the world population)and2.5millionnewcaseseachyear.Itsprevalencehas abimodaldistributionwithpeaksininfancyandolderage(65+).
For25–30%of allpatients, nocombination ofstandard therapy (medicationorsurgery)cancontroltheirseizures.Thesepatients aresaidtosufferfromrefractoryorintractableepilepsy.1–5
Compared to a normal population, the mortality risk for patientswithepilepsyiselevatedbyafactoroftwotothree,witha meanannualmortalityrateof1%.Epilepsy-relateddeathscanbe divided into four categories. First, deaths resulting from the underlyingcauseofepilepsy,neurologicaloranatomical(e.g.brain tumour).Second,deathscauseddirectlybyepilepsyitself(status epilepticusorSE,drowning,burning,trafficandothertraumatic accidents) oritstreatment.Third,deathsduetoco-morbidities, including depression leading to suicide, co-existent neurologic compromise,and respiratory tractinfectionorpneumoniaafter repetitiveaspirationduringseizures.Fourth,deathsofunknown cause,calledsuddenunexpecteddeathinepilepsy(SUDEP).6–8
SUDEP is defined as a sudden, unexpected, witnessed or unwitnessed,non-traumaticandnon-drowningdeath,occurring inbenigncircumstances,inanindividualwithepilepsy, withor withoutevidencefora seizure(withonsetwithinthepreceding hour)andexcludingdocumentedSE(seizureduration30minor Seizure22(2013)345–355
ARTICLE INFO
Articlehistory:
Received19October2012
Receivedinrevisedform14February2013 Accepted16February2013
Keywords:
Epilepsy SUDEP
Suddenunexpecteddeath Non-EEGseizuredetection Alarmsystem
ABSTRACT
Purpose:Thereisaneedforaseizure-detectionsystemthatcanbeusedlong-termandinhomesituations forearlyinterventionandpreventionofseizurerelatedsideeffectsincludingSUDEP(suddenunexpected deathinepilepticpatients).Thegold standardformonitoringepilepticseizuresinvolvesvideo/EEG (electro-encephalography),whichisuncomfortableforthepatient,asEEGelectrodesareattachedtothe scalp.EEGanalysis isalsolabour-intensive andhasyetto beautomatedandadaptedforreal-time monitoring.Itisthereforeusuallyperformedinahospitalsetting,forafewdaysatthemost.Thegoalof thisarticleistoprovideanoverviewofbodysignalsthatcanbemeasured,alongwithcorresponding methods,state-of-artresearch,andcommerciallyavailablesystems,aswellastostresstheimportanceof agooddetectionsystem.
Method: Narrativeliteraturereview.
Results: Arangeofbodysignalscanbemonitoredforthepurposeofseizuredetection.Itisparticularly interesting to include monitoring of autonomic dysfunction, as this may be animportant patho- physiologicalmechanismofSUDEP,andofmovement,asmanyseizureshaveamotorcomponent.
Conclusion: Themosteffectiveseizuredetectionsystemsaremultimodal.Suchsystemsshouldalsobe comfortableandlow-power.Thebodysignalsandmodalitiesonwhichasystemisbasedshouldtake accountoftheuser’sseizuretypesandpersonalpreferences.
ß2013BritishEpilepsyAssociation.PublishedbyElsevierLtd.Allrightsreserved.
*Correspondingauthor.Tel.:+3238215140.
E-mailaddresses:anouk.van.de.vel@uza.be(A.VandeVel),kris.cuppens@khk.be (K.Cuppens),bert.bonroy@khk.be(B.Bonroy),Milica.Milosevic@esat.kuleuven.be (M.Milosevic),katrien.jansen@uzleuven.be(K.Jansen),
Sabine.VanHuffel@esat.kuleuven.be(S.VanHuffel),
bart.vanrumste@esat.kuleuven.be(B.Vanrumste),lieven.lagae@uzleuven.be (L.Lagae),berten.ceulemans@uza.be(B.Ceulemans).
ContentslistsavailableatSciVerseScienceDirect
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/$–seefrontmatterß2013BritishEpilepsyAssociation.PublishedbyElsevierLtd.Allrightsreserved.
http://dx.doi.org/10.1016/j.seizure.2013.02.012
seizures without recovery in between), in which postmortem examinationdoesnotrevealacauseofdeath.9
Threeofthesemostcommoncausesofdeathcouldpossiblybe preventedthroughseizuredetection.SEcontributestoupto16%of deaths, traumatic accidents comprise up to 30% and SUDEP accountsforupto38%.1,10
Ofthelimitednumber(7–38%)ofwitnessedincidentsofSUDEP, 33–100%wereprecededbyseizures(inmostcases,convulsive).In theunwitnessedincidentsofSUDEP,itisnotalwaysclearwhether a seizure took place, although almost 90% involve evidence of tonguebiting, incontinenceor disrupted environment, strongly suggestingarecent convulsiveseizure(butnotexcludingother explanations).11–13
Thisarticleisbasedonourownwork,thedifficultiesthathave beenencounteredbyotherresearchteamsin thefield and the many requestsfromparents and patients for a robust seizure- detectionsystem,andinspiredbyourobservationthatinformation onthistopicisdifficulttoobtain.Eventhoughnon-EEG(electro- encephalography)systemshaveappearedonthemarketrecently, experiencewiththesesystemsisstilllimited.4,14–18Thegoalofthis articleistoprovidehealthcareprofessionals,engineers,aswellas interestedpatientswithanoverviewofbodysignalsthatcanbe measured, along withthe corresponding methods of measure- ment,state-of-artresearchandavailablesystems.Thearticlealso stressestheimportanceofagoodsystemforearlyintervention, reassuranceofpatientandenvironment,andpreventionofSUDEP.
2. SUDEP
2.1. Riskfactorsandtriggers
IthasbeensuggestedthatSUDEPiscausedbytheco-existence of several predisposing (inter-ictal) and triggering (peri-ictal) factors.The formerare known as risk factorsand thelatter as patho-physiologicalmechanismsordirectcauses. Knowledgeof thedifferentrisk factorscouldprovidecluestopossiblepatho- physiological mechanisms.Likewise, understanding themecha- nismsmayleadtotheidentificationofpreviouslyunrecognised riskfactorsthatcouldpossibly beamenabletocorrection, thus preventingSUDEP.8,19
Cardiac, respiratory and otherautonomic dysfunctions have beenthoroughlyinvestigatedandproposedaspatho-physiological mechanismsfor SUDEP. Thesedysfunctions result in decreased cerebraloxygen supplyandelectro-cerebralshutdown.Inother words,inhibitionofbrainactivitycanbebothacause(seizure)and aresponse.20
Risk factors that have been consistently identified include epilepsyonsetatyoungage,longdurationofepilepsy,generalised tonic–clonic seizures (GTCS), partial seizures (symptomatic epilepsy) and polytherapy.21 Furthermore, many risk factors constitute a risk due to their negative effects on the patho- physiologicalmechanisms.Forexample,youngpeopleareknown tohaveahigherbaselinevagaltone.22Inmanycases,GTCSpresent withtachycardiaand tachypnoea.23,24 Partialseizuresinfluence the cardiopulmonary instability: ictal asystole and apnoea or hypoventilationcanoccur intemporallobe seizures,while ictal tachycardiaand irregularbreathing often occur in hypermotor frontallobe seizures.25–27 Polytherapy can lead to cardiac and respiratory dysfunction.28 In other words, it is particularly interestingtoincludemethodsformonitoringcardiac,respiratory andotherautonomicbodysignalsindetectionsystems.
2.2. Prevention
Becausewearefarfromhavingaperfectmethodforpredicting seizures,letaloneSUDEP,supervisionisoneofthekeymeasuresin
addition to control over risk factors. In one large case–control study,supervisionwasassociatedwithasubstantialdecreasein theriskforSUDEP.11SupervisionmightthuspreventSUDEPboth by detectingseizures and by monitoring SUDEP risk factors or patho-physiological mechanisms, depending upon the methods used(Section4).
Possibleactionsbythecaregiverafteranalertincludeenabling aproperposition(fromproneorsupinetorecovery);stimulating the patient by rolling over or similar means (passive muscle movement); cardio-pulmonary resuscitation and defibrillation;
administrating medication or oxygen; clearing the airway;
protectingagainstinjuries;andseekinghelpifneeded.Supervision alsoallowsthecaregivertoreassureandcalmdownthepatient,to maintainavigilinthepost-ictalphaseandtocleanthebedincase ofvomitingorincontinence.
3. Seizuredetection
3.1. Seizuredetectioningeneral
Thegold standard formonitoring epilepticseizuresinvolves video/EEG(electro-encephalography),whichisuncomfortablefor the patient, as EEG electrodes are attached to the scalp. EEG analysisisalsolabour-intensiveandhasyettobeautomatedor adaptedforreal-timemonitoring.Itisthereforeusuallyperformed inahospitalsetting,forafewdaysatthemost,anditisgenerally usedtoconfirmthediagnosisofepilepsyinsteadofforobjectively quantifyingseizurefrequencyorwarningforpotentialSUDEP.
Theinterestinambulantsystemsoriginatesfromtheneedto monitorpatientsoverextensiveperiods(weekstomonths)andin the natural home and outdoor environment (domiciliary and residential settings).29 Long-term home monitoring might also reducecostsbyeliminatingtheneedforhospitaladmissionand clinicalobservation.
Detectingseizuresmakesitpossibletoalertthecaregiver,who cantakeaction,asdescribedinSection2.2.Generatinganobjective logofthenumber,order,duration,affectedbodypartandtypeof seizures can inform the neurologist and help to optimise treatment.Objectiveseizurelogscouldsignificantlyenhancethe somewhatunreliablepatient-diarytechnique,whichissubjective and can beinfluenced by peri-ictally affectedconsciousness or memory. Onelimitationof non-EEGseizure detectionis thatit cannotprovideretroactiveverification(byvideo/EEG)forevents thatareloggedasseizures.30
Itshouldbemadepossibletoadjustseizurealertsaccordingto theuser’spersonalpreferences.Some parentsmightwanttobe informedofeveryseizure,inordertoreassuretheirchild,while institutionsmightrequestthisinformationinordertocomplywith insurancerequirements.Otherparentsmightnotwanttowakethe patientunnecessarily,andalsohospitalcaregiversmightpreferto be informedonly in the eventof life-threatening seizuresthat couldsignalnear-SUDEPeventsandrequireimmediateaction.In otherwords,anidealsystemshouldbecustomisedtoseizuretype, aswellastoseizureseverityandqualityoflife.
Anotherfactortoconsideriswhicheventsneeddetection.One commondifficultyinvolvesdistinguishingepilepticseizuresfrom non-epileptic events (e.g. syncope, hypoglycaemic seizures in diabetes,psychogenicnon-epilepticseizuresorpseudo-seizures), life-threatening events (e.g. suffocation) and sleep-related dis- orders (e.g. restless-leg syndrome, periodic limb movement disorder,somnambulism,pavornocturnus).Howevertherelative needfor differentialdiagnosisis unclear.Becausenon-epileptic events mightalsoneedcare, similarsafetyprotocols shouldbe administered.31 Furthermore, many patients have overlapping disorders.Forexample,athirdofallpatientswithnocturnalfrontal lobe seizures (NFLS) have a history of somnambulism, pavor
nocturnus,rhythmicmovementdisorderand enuresis.32On the other hand, detecting ‘normal’ sleep-related movement or vocalisations(e.g.nocturnallegcramps,hypnicjerks,somniloquy) mightberedundant.
Monitoringduringthedayandwhenthepatientisawakeis useful primarily for (young) adults. It is important when the patientisalone,andallowsforemergencycareincaseofdrowning, burning,trafficandothertraumaticaccidents.Oneadvantageof daytimemonitoringisthatfalsealarmsarelessproblematicand canbe turnedoff bythepatient. On theotherhand, detection during sleep might be easier, as patients remain at the same locationwithinthedetectionrange,sincethereislessnoisefrom voluntary movements and because convulsions occur in a relatively controlled, reproducible manner. It is also more important because of the reduction or lack of supervision, as subjective elements that precede seizures (auras that allow patientstoforeseethem)are unavailableandsince sometypes (particularlyfrontallobeseizures)33occurmoreoftenatnight.
3.2. Intracerebralseizuredetection
The main reasons why many studies focus on the early detection or prediction of seizures using EEG (cortical or intracranial)areobviouslythatepilepsyisaneurologicaldisease andthattheultimatepurposeshouldbetopreventdamagetothe brainbypreventingseizures.
Electrographicseizurescanalreadystartbeforeandpropagate totheclinicallyobservableseizure,thusallowingtheprediction and possible prevention of clinical seizuresor SUDEP (Fig. 1).
Automated EEG analysis could also allow predicting/detecting seizures that remain sub-clinical or those for which clinical expressionisverysubtle(e.g.absenceseizures).
Inadditiontotherisksofimplantingintracranialsystemsand thediscomfortofhead-attacheddevices,theyraiseanumberof otherquestionsaswell.
DoesEEGdetectallseizures?Somenocturnalseizuresproduce subtleornoictalEEGpatternsduringscalprecording,ortheirEEG patternsmaybeobscuredbyexcessivemuscleartefacts.Infact, certain forms of epilepsy (in which seizures occur almost exclusively at night, including NFLS, where ictal EEG findings are visible in only half of all cases)34 were not definitively recognisedasepilepticuntilthepast20–25years.
ShouldEEGrecordingsbecombinedwithphysiologicalsources asastandardform,giventhatthedifficultyofdetectingseizuresis exacerbated by contaminationof EEGrecordings withenviron- mentalandbiologicalartefacts?35
Couldaseizureindeedbeabortedthroughstimulationafterthe onsetofelectrographicseizureactivity,orhasthebrainalready passed the ‘point of no return’, to a state that will inevitably progressintoaclinicalseizuremanifestation?36Inotherwords, how important is the latency period between the onset and detection of electrographic seizures for allowing alarming and action?
Dosub-clinicaleventsrequireactionatall?Whowouldbenefit frompreventive measures?Whichactionsshouldbetaken and when,bothforpreventingseizuresandforpreventingSUDEP?
3.3. Extracerebralornon-EEGseizuredetection
Mostengineersandneuroscientistsfocusonthelinkbetween brainandepilepsyandnotonthelinkbetweenbrainandbody.
Extracerebralbodysignalsareundercorticalcontrolandalteredby seizuresaswell,thusprovidingindirectbutvaluableinformation aboutthestateofthebrain.30
Given that no ideal intracerebral devices are yet available (automated detection and predictability of seizures have been studiedsincethe1970s)36andconsideringthemanyadvancesof non-EEGseizuredetectiondescribedbelow,itis surprisingthat such devices are not mentioned more often in strategies for epilepsyandpreventingSUDEP.
First, regardless of their value, scientific advances may not translate into improved care unless the devices are broadly accessible.Extracerebral(or,morespecifically,non-EEGdevices) aremorewidelyapplicablethanintracranialdevicesare,asthey arenotinvasiveandcan,ifcontactless,evenbenon-obtrusive.30In othermedicalconditions,manydevicesforphysiologicalsigns(e.g.
forsleepdisorders)andmovementdetection(e.g.forParkinson andfalldetection)arealreadyon themarket,thusallowingthe transfer of knowledge. Non-EEG body signals might be less complex.30Theymightthereforeallowsmallerandlessexpensive devicesthatuselesspowerforautomatedanalysis.Finally,non- EEGdevicescouldimmediatelybetestedonhumansubjects.
Similar to coupling intracerebral detection to a preventive or aborting action (Fig. 1), extracerebral or, more specifically, non-EEG devices could include responsive stimulation of heart, respiration or muscles (including diaphragmatic pacing), administration of medication or oxygen or (less obviously)the inflationofan‘airbag’topreventinjury.Thelatterhasbeentested in thecontextof hipprotectionin ordertopreventfall-related injuriesintheelderly.37Sostrategiesforepilepsyandpreventing SUDEPshouldnotfocussolelyon‘cure’(ifpreventingoraborting seizures can be considered a cure) but also on ‘care’ (disease control)aspartoftheoveralltreatmentplan.
4. Detectionsystems 4.1. Slowprogression
Despitethepleaforadditionalresearchfocusingonnon-EEG detectiondevicesforepilepticseizures,thefieldismovingslowly, andthereisnoclearoverviewofthedevicesthatareonthemarket.
Theobtrusivenessandexcessivefalsealarmsassociatedwiththese devices often make patients reluctant to use them. Moreover, many commercial systems detect primarily intense, long and repetitive movements in general, and there is apparently no scientificprooforinformationfromclinicaltrialstoprovideusers withreliableinformationontheirefficacy.
Perhapsthegreatestdifficultytoovercome,andonereasonwhy companiesmightbereluctanttoinvest,involvesthenecessityof customising the devices to the patient’s seizures (supervised learning or patient-specific training) and the user’s personal preferences(Section3.1).
Fig.1.EEGsampleofaseizure.Predictioncanleadtopreventionandearlydetection toabortionofseizures.
A.VandeVeletal./Seizure22(2013)345–355 347
