Applicationvalueofanewgenerationmodel-basediterativereconstructioninchestCTscan / 实用放射学杂志

ABSTRACT

Objective ToinvestigatetheimpactofCTimagequalityforfilteringbackprojection(FBP),conventionalmodel-based iterativereconstruction(MBIRC)andnewgeneration model-basediterativereconstruction (MBIRN)onchest.Methods Thirtypatientswith chestCTscanwerecollected.FBP,MBIRCandMBIRN wereusedtoreconstructtheimage.Objectivequality[standarddeviation(SD) valueoftheROI,SNR],thenoisereductionrateandSNRimprovementrateofMBIRCand MBIRN withrespecttoFBP werecom-paredacrossthethreeimages.Atthesametime,tworadiologistsusedtheblind methodtoevaluatetheintrapulmonarystructurein thelungalgorithm FBP,MBIRC,MBIRN,andthemediastinalstructure (5-pointsystem)inthestandardalgorithmsFBP,MBIRC, MBIRN.Results ComparedwithFBP,theimagemusclenoisesofMBIRCand MBIRN were76.71% and86.06%lowerthanFBP,respectively, andthefatnoiseswere66.91% and78.18%lowerthanFBP,respectively.Thedifferencewasstatisticallysignificant(P<0.05).The imageSNRofMBIRCandMBIRN were74.12% and84.97% higherthanthatoftheFBPgroup,respectively.ThefatSNRwere65.63% and 76.02% higherthanthatoftheFBPgroup (P<0.05).Thethreealgorithmsshowedstatisticallysignificantdifferencesinsubjective noise,intrapulmonaryvascular,bronchialresolution,mediastinalbloodvessels,andlymphnodes.MBIRN hadthelowestsubjective noise,andthehighestSNR,mediastinalstructure,andintrapulmonaryvesselsandbronchi.Conclusion Comparedwith MBIRC and FBPwithnormaldosechestCTscan,MBIRN cansignificantlyreducethenoiseofchestCTscanimages,improveSNR,and more clearlyshowthedetailsofthescanrangeandlesionedgefeatures.