论文部分内容阅读
Purpose:ToevaluatethediagnosticefficacyofneuriteorientationdispersionanddensityImaging(NODDI)anddiffusion kurtosisimaging(DKI)ingradinggliomas. Methods and Materials:29patients(male:18,female:11,meanage:45.4y)wereprospectivelyrecruitedandtheyunderwent conventional,whole-braindiffusion-weightedimageswhichwerecollectedatthreebvalue(0,1000and2000s/mm2)andeach non-zerobvaluehas30directions.Boththeb=1000s/mm2andb=2000s/mm2datawereusedfortheNODDIandDKI analysis.kurtosis(MK),axialkurtosis(Ka),radialkurtosis(Kr),meandiffusivity(MD),fractionalanisotropy(FA),apparentdiffusion coefficient(ADC),Neuritevolumefraction(ficvf)andorientationdispersionindex(ODI)mapsweregenerated.TensameROIs weremanuallydrawnoneachmapsinthetumorparenchymabyImageJandthemeanvalueofeachparameterswererecorded. Mann-Whitney-Wilcoxontest,linearregressionandROCanalysiswereperformed. Results:AlldiffusionparameterscansignificantlydifferentiateHGGandLGG(p<0.000).MK,Ka,Kr,FA,ficvfandODIwere significantlyhigherinHGGwhileMDandADCweresignificantlylowerinHGG.ROCanalysisshowedthatficvfhadthehighest diagnosticvalue(AUC:0.80,sensitivity:71%;specificity:80%;cut-offvalue:0.33)inpredictingHGGwhileADCdemonstratedthe lowestdiagnosticvalue.CombininganalysisofADC,MKandficvfbylinearregressionshowedthattheAUCforpredictingLGG wasmuchhigher(89%)andthesensitivityandspecificitywere77%and88%,respectively. Conclusion:Advanceddiffusionweightedimaging(NODDIandDKI)mayhelpingliomasgradingandficvfshowedthehighest diagnosticvalue.Combininguseofdifferentdiffusionmodelscouldsubstantiallyimproveourdiagnosis.