不同MRI序列对脑室炎的探测比较（英）（1）

ComparisonofMRISequencestoDetectVentriculitis

AkiraFujikawa1,KazuhiroTsuchiya2,KeitaHonya2,ToshiakiNitatori2

1DepartmentofRadiology,JapanSelf-DefenseForcesCentralHospital,1-2-24,Ikejiri,Setagaya,Tokyo154-8532,Japan.AddresscorrespondencetoA.Fujikawa.

2DepartmentofRadiology,KyorinUniversityHospitalSchoolofMedicine,Tokyo,Japan.

ReceivedDecember18,2004;acceptedafterrevisionAugust22,2005.

AJROctober2006;187:10481053,AmericanRoentgenRaySociety

Keywords:brain,diffusion-weightedimaging,FLAIRimaging,infectiousdiseases,MRI

OBJECTIVE.TheabilityofdifferentMRIsequencestodepictcharacteristicfindingssuggestiveofventriculitiswascompared.

CONCLUSION.Thestudycomprised20brainMRIstudiesin13patientswhohadafinaldiagnosisofventriculitis.Bothdiffusion-weightedimagingandFLAIRimagingwereequallyandhighlysensitivefordetectingintraventriculardebrisandpusthemostcommonMRIfindingsuggestiveofventriculitis.FLAIRimagingwassuperiortocontrast-enhancedT1-weightedimagingfordepictingventricularwallabnormalitiesalesscommonfindingthatalsoissuggestiveofventriculitis.

VentriculitisisanuncommonCNSinfectionthathasbeendescribedusingavarietyoftermsincludingependymitis,intraventricularabscess,ventricularempyema,andpyocephalus[1].Thisvarietyoftermsreflectsvariousfacetsofthedisease’spathologicprocess.Becauseventriculitisisasevereintracranialinfectionthatcanleadtoserioussequelaeanddeath,promptdiagnosisisnecessary.However,theclinicalfeaturesofventriculitisareoftenobscureandnonspecific.MRIplaysanimportantroleasafirst-linediagnostictoolinthediagnosisofventriculitis[2].

PreviouslyreportedcharacteristicMRIfindingsofventriculitisincludeintraventriculardebrisandpus,abnormalperiventricularandsubependymalsignalintensity,andenhancementoftheventricularliningonconventionalMRIsequences[3,4].Inaddition,afewreportshaveillustratedtheusefulnessofdiffusion-weightedimagingfordetectingintraventriculardebrisandpus[1,3,5].WehaveencounteredcasesofventriculitisinwhichtheaboveMRIfeaturesweresubtleonsomeMRpulsesequencesbutobviousonothers.Thus,knowingwhichMRIsequencesareusefulfordetectingthecharacteristicfindingsofventriculitisisimportantfordailyclinicalpractice.

Tothebestofourknowledge,theutilityofdifferentMRIsequencesfordetectingthecharacteristicfindingsofventriculitishasnotbeencompared.Inthisstudy,wesoughttodeterminewhichcombinationsofMRIsequencesandfindingswereusefulforthediagnosisofventriculitis.

MaterialsandMethods

Bysearchingthecomputerdatabaseofourinstitution’sdepartmentofradiology,weidentified13patientswhohadbeentreatedbetweenNovember1999andJuly2004andwerestronglysuspectedofhavingventriculitisbasedontheirMRIfindings.Themedicalrecordsofthesepatientswerereviewedbytwooftheauthorstoobtainclinicalfollow-upinformationincludinglaboratory,treatment,andoutcomedataandtoconfirmthefinaldiagnosisbasedontheradiologicandclinicalfindings.

Thediagnosesofthe13patientswereconsistentwithventriculitis.Thepatients,eightmalesandfivefemales,hadameanageof50.5years（agerange,newborn85years）.ThepathogensdetectedwereKlebsiellapneumoniae（n=1）,Staphylococcusaureus（n=1）,Streptococcuspneumoniae（n=2）,Enterobactercloacae（n=1）,Enterobacteraerogenes（n=1）,Escherichiacoli（n=1）,Mycobacteriumtuberculosis（n=1）,Pseudomonasaeruginosa（n=2）,Cryptococcusneoformans（n=1）,andunknown（n=2）.ThespecimensforculturesandGramstainswereobtainedbylumbarpuncture（n=9）,ventriculostomytube（n=3）,andabscessdrainage（n=1）.Threepatientswereclinicallyimproved.Fivepatientshadprolongeddisturbanceofconsciousness.Fivepatientsdiedduringthestudy’stimecourse（273daysafterMRI）.Underlyingconditionsofthesepatientsincludedmyelodysplasticsyndrome（n=1）,lymphocyticleukemia（n=1）,postclippingofcerebralaneurysm（n=1）,malignantlymphoma（n=1）,pneumonia（n=1）,severefacialbonefracture（n=1）,AIDS（n=2）,mastoiditisandpetrositis（n=1）,postresectionofmeningioma（n=1）,andcerebellarabscess（n=1）.Twopatientshadnoriskfactor.Atotalof20brainMRIstudiesperformedinthese13patientswereincludedintheimaginganalyses.

AllMRIwasperformedusingoneofthree1.5Timagingsystems（VisartorExelart,ToshibaMedicalSystems;GyroscanIntera,PhilipsMedicalSystems）.AllMRIstudieswereperformedaccordingtoourinstitution’sprotocolforbraininfectionandinflammation,althoughtheimagingparametersvariedslightlydependingonthemanufacturerofthesystem.TheMRIprotocolincludedaxialfastspin-echoT2-weightedsequences（TR/TErange,4,0004,900/90120;fieldofview[FOV],1822×22cm;matrix,160192×256384;numberofexcitations[NEX],12;echo-trainlength,13,15,or17;receiverbandwidth,39.68,41.856,or62.464kHz）;axialcontrast-enhancedspin-echoT1-weightedsequences（450540/1015;FOV,1822×22cm;matrix,160176×258384;NEX,12）;axialFLAIRsequences（8,00010,000/105120;FOV,18×22;matrix,160192×256320;NEX,12;inversiontime,2,3002,600msec）;andaxialsingle-shotspin-echoecho-planardiffusion-weightedsequenceswithbvaluesof0and1,000s/mm2alongallthreeorthogonalaxes（4,0008,000/95120;FOV,2225×2630cm;matrix,128×128;NEX,1）.Calculatedapparentdiffusioncoefficient（ADC）mapswerealsoobtained.

Threeradiologistswithexperienceinbrainimagingretrospectivelyreviewedatotalof100hard-copyMRimages.Hard-copyimagesofeachsequencewereseparatedfromeachotherandrandomlymixed.Eachradiologistindependentlyassessedthefollowingfindings:thepresenceofabnormalintraventricularsignalintensity,thepresenceofabnormalperiventricularsignalintensity,andthepresenceofcontrastenhancementoftheventricularwalloncontrast-enhancedT1-weightedimages.ThepresenceofotherMRIfindingsassociatedwithintracranialinflammation,includinghydrocephalus,meningitis,brainabscess,cerebritis,andchoroidplexitis,wasalsoassessed.Abnormalintensitiesonthediffusion-weightedimageswerecorrelatedwithcorrespondingfindingsontheADCmaps.TheADCvaluesofabnormalintraventricularintensitiesandnormal-appearingintraventricularCSFwereobtainedbycalculatingthemeanvaluesofthreecircularregionsofinterest（ROIs）,eachofwhichwas25mmindiameter,ontheADCmaps.ForabnormalperiventricularintensitiesonFLAIRimages,eachreviewerattemptedtosubjectivelydistinguishbetweenthepresenceofaninflammatoryprocessandanage-relatednormalvariant.Interpretationdiscrepancieswereresolvedbythejudgmentofathirdradiologist.

Wefollowedourinstitution’sethicalguidelines.Atourinstitution,institutionalreviewboardapprovalandinformedconsentarenotrequiredforretrospectivereviewsofimagingstudiesandmedicalrecords.

Ofthe20MRIstudies,intraventricularabnormalintensitieswerepresentin19（95%）ofthediffusion-weightedimages,19（95%）oftheFLAIRimages,13（65%）oftheT2-weightedimages,and10（50%）ofthecontrast-enhancedT1-weightedimages.OfallMRIstudies,abnormalintensitiesweredetectedinbilateralventriclesin12ofthediffusion-weightedimages,nineoftheFLAIRimages,fiveoftheT2-weightedimages,andfiveofthecontrast-enhancedT1-weightedimages.AbnormalintraventricularsignalintensitiesrelativetothesignalintensitiesfornormalCSFwerehyperintenseonthediffusion-weightedandFLAIRimages,slightlyhypointenseontheT2-weightedimages,andslightlyhyperintenseonthecontrast-enhancedT1-weightedimages（Fig.1）.

Fig.168-year-oldmanwithseveremastoiditisandacutepetrositis.

A,T2-weightedimageshowsareasofslighthypointensityrelativetoCSFinbilateraltrigoneoflateralventricle.Thisfindingissuggestiveofintraventriculardebrisandpus.Slightventricularwallabnormalityisnoted.

B,FLAIRimageshowshyperintenseintraventricularlesionsrelativetoCSFandhyperintensityalongventricularlining,suggestingependymitis.

C,Contrast-enhancedT1-weightedimageshowsabnormalcurvilinearenhancementalongtheventricularwall.IntraventriculardebrisandpusareslightlyhyperintenserelativetoCSF.

D,Diffusion-weightedimageshowsareasofconspicuousintraventricularandperiventricularhyperintensity,indicatingrestrictedwaterdiffusioninthoseareas.

E,Apparentdiffusioncoefficient（ADC）mapshowsareasofdecreasedADCvaluesincorresponding