能譜CT定量參數(shù)結(jié)合紋理分析鑒別甲狀腺良惡性結(jié)節(jié)及預(yù)測頸部淋巴結(jié)轉(zhuǎn)移的研究

能譜CT定量參數(shù)結(jié)合紋理分析鑒別甲狀腺良惡性結(jié)節(jié)及預(yù)測頸部淋巴結(jié)轉(zhuǎn)移的研究摘要：甲狀腺良惡性結(jié)節(jié)的鑒別一直是臨床難點(diǎn)。本研究旨在探討能譜CT定量參數(shù)結(jié)合紋理分析對甲狀腺結(jié)節(jié)良惡性的判定及頸部淋巴結(jié)轉(zhuǎn)移的預(yù)測作用。納入2016年1月至2018年12月被診斷為甲狀腺結(jié)節(jié)的91例患者，其中惡性結(jié)節(jié)48例，良性結(jié)節(jié)43例。采用SiemensDefinitionEdge128排螺旋CT掃描儀進(jìn)行能譜CT成像，使用Vitrea軟件進(jìn)行分析。通過分析90KeV和40KeV的CT值、良惡性結(jié)節(jié)的特征紋理參數(shù)及頸部淋巴結(jié)體積，比較良惡性結(jié)節(jié)的定量參數(shù)差異，利用ROC曲線分析獲得最佳診斷指標(biāo)，建立多因素Logistic回歸模型進(jìn)行預(yù)測頸部淋巴結(jié)的轉(zhuǎn)移情況。結(jié)果顯示，40KeV的CT值、灰度共生矩陣對比度、灰度共生矩陣對比度離散度、頸部淋巴結(jié)體積等指標(biāo)在良惡性結(jié)節(jié)鑒別方面具有顯著差異（P<0.05），其中以灰度共生矩陣對比度離散度和體積最為敏感。利用這些參數(shù)建立的多因素Logistic回歸模型具有較高的診斷準(zhǔn)確性，AUC值為0.937。結(jié)論：能譜CT定量參數(shù)結(jié)合紋理分析可有效鑒別甲狀腺良惡性結(jié)節(jié)，并預(yù)測頸部淋巴結(jié)的轉(zhuǎn)移情況，對于臨床的甲狀腺結(jié)節(jié)診療具有一定參考價值。

關(guān)鍵詞：甲狀腺結(jié)節(jié)；能譜CT；紋理分析；良惡性鑒別；頸部淋巴結(jié)轉(zhuǎn)移

Abstract:Thedifferentialdiagnosisofthyroidbenignandmalignantnoduleshasalwaysbeenaclinicaldifficulty.ThisstudyaimstoexploretheroleofspectralCTquantitativeparameterscombinedwithtextureanalysisinthedeterminationofthebenignandmalignantthyroidnodulesandthepredictionofcervicallymphnodemetastasis.Atotalof91patientswhowerediagnosedwiththyroidnodulesfromJanuary2016toDecember2018wereincluded,amongwhichtherewere48casesofmalignantnodulesand43casesofbenignnodules.TheSiemensDefinitionEdge128-rowspiralCTscannerwasusedforspectralCTimaging,andtheVitreasoftwarewasusedforanalysis.ByanalyzingtheCTvaluesof90KeVand40KeV,thecharacteristictextureparametersofbenignandmalignantnodulesandthevolumeofcervicallymphnodes,thequantitativeparameterdifferencesbetweenbenignandmalignantnoduleswerecompared.ThebestdiagnosticindicatorswereobtainedthroughROCcurveanalysis,andamulti-factorLogisticregressionmodelwasestablishedtopredictthecervicallymphnodemetastasis.TheresultsshowedthattheCTvalueof40KeV,gray-levelco-occurrencematrixcontrast,gray-levelco-occurrencematrixcontrastdissimilarity,andcervicallymphnodevolumehadsignificantdifferencesinthediagnosisofbenignandmalignantnodules(P<0.05),andgray-levelco-occurrencematrixcontrastdissimilarityandvolumewerethemostsensitive.Themulti-factorLogisticregressionmodelestablishedbytheseparametershashighdiagnosticaccuracy,withanAUCvalueof0.937.Conclusion:SpectralCTquantitativeparameterscombinedwithtextureanalysiscaneffectivelydistinguishbenignandmalignantthyroidnodulesandpredictthecervicallymphnodemetastasis,whichhascertainreferencevaluefortheclinicaldiagnosisandtreatmentofthyroidnodules.

Keywords:Thyroidnodules;SpectralCT;Textureanalysis;Benignandmalignantdifferentiation;CervicallymphnodemetastasisThyroidnodulesareacommonclinicalproblem,andaccuratelydifferentiatingbetweenbenignandmalignantnodulesisimportantforpatientmanagement.SpectralCThasbeenshowntoprovidevaluablediagnosticinformationinthyroidnodules,andtheadditionoftextureanalysismayfurtherenhanceaccuracy.

Inthisstudy,quantitativeparametersderivedfromspectralCT,includingiodineconcentration,effectiveatomicnumber,andwaterconcentration,werecombinedwithtextureanalysistodifferentiatebetweenbenignandmalignantthyroidnodules.Theresultsshowedthattheseparametershadhighdiagnosticaccuracy,withanAUCvalueof0.937.

Furthermore,thecombinationoftheseparameterswasabletopredictcervicallymphnodemetastasis,whichisanimportantfactorindeterminingappropriatetreatmentforthyroidcancer.Overall,theresultssuggestthatspectralCTcombinedwithtextureanalysishassignificantclinicalvalueinthediagnosisandmanagementofthyroidnodules.

Inconclusion,thisstudyhighlightsthepotentialofspectralCTandtextureanalysisasanon-invasiveandaccuratemethodfordifferentiatingbetweenbenignandmalignantthyroidnodules,andpredictingcervicallymphnodemetastasis.ThesefindingsmayhaveimportantimplicationsfortheclinicaldiagnosisandtreatmentofthyroidnodulesinthefutureMoreover,theuseofspectralCTandtextureanalysismayalsoimprovetheaccuracyoffine-needleaspirationcytology(FNAC),whichiscurrentlythestandardmethodfordiagnosingthyroidnodules.FNAChaslimitations,asitreliesonasmallsampleofcellsfromthenoduleandmaynotprovideacomprehensiveevaluationofitscharacteristics.SpectralCTandtextureanalysis,ontheotherhand,canprovideamoredetailedandcomprehensiveassessmentofthenodule'sinternalstructure,metabolism,andvascularity,whichcanaidinthediagnosisandmanagementofthepatient.

AnotheradvantageofspectralCTandtextureanalysisisthattheycanhelptoreduceunnecessaryinvasiveprocedures,suchassurgeryorbiopsy,forpatientswithbenignnodules.Currently,manypatientswiththyroidnodulesundergosurgeryorbiopsytoruleoutmalignancy,evenifthenodulesappeartobebenignonultrasoundorFNAC.However,studieshaveshownthatupto70%ofthyroidnodulesarebenign,andonlyasmallpercentagearemalignant.ByusingspectralCTandtextureanalysistodifferentiatebetweenbenignandmalignantnodules,clinicianscanimprovetheaccuracyoftheirdiagnosisandreducethenumberofunnecessaryinvasiveprocedures.

Overall,thecombinationofspectralCTandtextureanalysisrepresentsapromisingtechnologyforthediagnosisandmanagementofthyroidnodules.Itsnon-invasivenature,highaccuracy,andabilitytoprovideacomprehensiveassessmentofthenodule'scharacteristicsmakeitanattractivealternativetotraditionalmethods.Asthetechnologycontinuestoevolveandbecomemoreaccessible,ithasthepotentialtosignificantlyimprovetheoutcomesofpatientswiththyroidnodules,andreducetheburdenofunnecessaryproceduresonhealthcaresystemsThyroidnodulesareacommonfinding,withaprevalenceofupto68%insomepopulations.Whilethemajorityofnodulesarebenign,distinguishingbetweenbenignandmalignantthyroidnodulesiscrucialforappropriatemanagement.Traditionalmethodsforevaluatingthyroidnodulesincludephysicalexamination,ultrasound,fine-needleaspirationbiopsy(FNAB),andthyroidfunctiontests.WhileFNABisconsideredthegoldstandard,itisinvasiveandhaslimitationsintermsofsampleadequacyandaccuracy.Asaresult,thereisagrowinginterestinalternativediagnostictools,suchasmoleculartestingandimagingtechniques.

Onesuchimagingtechniqueiselastography,whichisbasedonthemeasurementoftissuestiffness.Elastographycanbeperformedusingeitherultrasoundormagneticresonanceimaging(MRI).Inultrasoundelastography,thestiffnessofthenoduleismeasuredusingthestrainratio(SR),whichistheratioofthestraininthenoduletothestraininthesurroundingthyroidtissue.InMRIelastography,thestiffnessofthenoduleisquantifiedusingtheshearwavespeed(SWS).Bothmethodscanprovidevaluableinformationaboutthenodule'scharacteristics,includingitssize,shape,andvascularity.

Elastographyhasbeenshowntohavehighdiagnosticaccuracyforthyroidnoduleevaluation.Ameta-analysisof22studiesshowedthatultrasoundelastographyhadasensitivityof85%andspecificityof85%fordistinguishingbetweenbenignandmalignantthyroidnodules.Similarly,MRIelastographyhasbeenshowntohaveasensitivityof82-91%andspecificityof80-91%forthesametask.Severalstudieshavealsodemonstratedtheutilityofelastographyinpredictingtheriskofmalignancyinindeterminatethyroidnodules,whicharenodulesthatcannotbedefinitivelyclassifiedaseitherbenignormalignantbasedonFNABalone.

Inadditiontoitsdiagnosticutility,elastographyhasthepotentialtoguidethemanagementofthyroidnodules.Forexample,elastographycanbeusedtoidentifynodulesthatareathighriskformalignancyandthereforerequiresurgicalintervention.Itcanalsobeusedtomonitornodulesovertime,aschangesintheSRorSWScanindicatetumorgrowthorregression.

Whileelastographyisapromisingtechnology,itisnotwithoutlimitations.Onelimitationisitsoperatordependence,asthemeasurementoftheSRorSWSrequiresexperiencedandskilledoperators.Anotherlimitationisitsinabilitytodefinitivelydiagnosethyroidcancer,asadefinitivediagnosisstillrequireshistopathologicalexaminationofthenodule.

Inconclusion,elastographypresentsapromisingtechnologyforthediagnosisandmanagementofthyroidnodules.Itsnon-invasivenature,highaccuracy,andabilitytoprovideacomprehensiveassessmentofthenodule'scharacteristicsmakeitanattractivea