双能CT定量参数联合淋巴结RADS评分预测头颈部鳞状细胞癌淋巴结转移  

Value of quantitative parameters of dual-energy CT combined with Node-RADS in prediction of lymphnode metastasis in HNSCC

作　　者：张丽萍[1,2] 苏童[1] 伍晓倩 曲江明 王天娇 徐振谭 陆晓平 陈钰 张竹花[1] 冯逢[1] 金征宇[1] ZHANG Li-ping;SU Tong;WU Xiao-qian(Department of Radiology,Peking Union Medical College Hospital,Chinese Academy of Medical Sciences&Peking Union Medical College,Beijing 100031,China)

机构地区：[1]中国医学科学院北京协和医学院北京协和医院放射科,北京100032 [2]中山市人民医院医学影像中心,广东中山528403

出　　处：《放射学实践》2025年第4期450-455,共6页Radiologic Practice

基　　金：国家自然科学基金(82371962);中央高水平医院临床科研业务费(2022-PUMCH-B-067);2021 SKY影像科研基金(Z-2014-07-2101)。

摘　　要：目的:探讨双能CT(DECT)定量参数联合淋巴结影像数据和报告系统(Node-RADS)评分术前预测头颈部鳞状细胞癌(HNSCC)淋巴结转移的价值。方法:回顾性将2016年6月-2024年6月在本院确诊为HNSCC的57例患者共176枚淋巴结纳入本研究。根据病理结果,分为转移淋巴结组(TL,n=100)和非转移淋巴结组(NTL,n=76)。测量每个淋巴结的DECT定量参数,包括虚拟平扫(VNC)CT值、碘浓度(IC)、脂肪分数(FF)、电子云密度(ED)、有效原子序数(Zeff)、标准化碘浓度(NIC)及单能级40keV(VMI)图像上的CT值,计算40~100 keV能谱曲线斜率λ和双能量参数(DEI)。基于Node-RADS 1.0对每个淋巴结进行Node-RADS评分。采用多因素logistic回归和LASSO回归进行变量筛选,建立基于DECT定量参数和Node-RADS评分的回归模型。利用受试者操作特征(ROC)曲线评估模型的预测效能,采用DeLong检验比较两个模型AUC的差异。结果:TL组和NTL组的RADS评分、Zeff、VMI 40keV的CT值、λ和DEI值的差异均有统计学意义(P<0.05)。经多因素logistic回归分析,构建的预测模型中纳入了2个特征,分别是VMI 40keV的CT值(OR=1.005,95%CI:1.001~1.009,P=0.02)和Node-RADS评分(OR=0.237,95%CI:0.1521~0.369,P>0.05);当VMI 40keV值≤255.99 HU,Node-RADS评分≥2分时,预测转移性淋巴结的AUC、敏感度、特异度和符合率分别为0.849(95%CI:0.794~0.904)、93.4%、62.0%和75.6%。经LASSO回归分析,模型评价指标最佳的Lambda值(lambda.min)为0.0146,筛选出的变量为DEI和Node-RADS评分;当DEI≤0.0461、Node-RADS评分≥3分时,预测转移性淋巴结的AUC、敏感度、特异度和符合率分别为0.850(95%CI:0.795~0.905)、93.4%、64.0%和76.8%。Logistic回归模型和LASSO回归模型预测转移性淋巴结的AUC的差异无统计学意义(Z=-0.546,P>0.05)。结论:DECT定量参数中淋巴结VMI 40keV的CT值、DEI分别联合Node-RADS评分构建的诊断模型,可用于预测HNSCC淋巴结转移。Objective:To investigate the value of quantitative parameters of dual-energy CT(DECT)combined with Node Reporting and Data System(Node-RADS)in prediction of lymph node metastasis in head and neck squamous cell carcinoma(HNSCC).Methods:This study was a retrospective study,in which 57 patients and 176 lymph nodes were enrolled from Jun 2016 to Jun 2024.The 176 lymph nodes were divided into metastatic group(TL,n=100)and non-metastatic group(NTL,n=76).The following DECT quantitative parameters of each lymph node were assessed:CT value on virtual non-contrast(VNC)images,iodine concentration value(IC),fat fraction(FF),electronic density(ED),effective atomic number(Zeff),standardized iodine concentration(NIC),CT value on virtual mono-energetic images(VMI),the slope of the 40~100keV spectral curve(λ)and dual-energy index(DEI).The Node-RADS score of each lymph node was recorded based on Node-RADS 1.0.Multivariate logistic regression analysis and least absolute shrinkage and selection operator(LASSO)regression were used to establish a regression model based on DECT quantitative parameters and Node-RADS score.The ROC curve was performed to evaluate the diagnostic performance of each model.DeLong test was used to compare the difference in area under the curve(AUC)of the quantitative parameters.Results:Statistically significant difference was found in Node-RADS score,Zeff,CT value on VMI 40keV,λand DEI between the metastatic and non-metastatic nodes(P<0.05).After a multivariate logistic regression analysis,2 features were incorporated in the model:VMI 40keV(OR=1.005,95%CI:1.001~1.009,P=0.02)and the Node-RADS score(OR=0.237,95%CI:0.1521~0.369,P>0.05),and the cut-off point was at VMI 40keV 255.99HU and Node-RADS 2 scores.The AUC,sensitivity,specificity,and accuracy of the predicting model were 0.849(95%CI:0.794~0.904),93.4%,62.0%and 75.6%,respectively.After LASSO regression analysis,the best lambda value of the model evaluation index(lambda.min)was 0.0146,the incorporated features were DEI and Node-RADS score.With the cutoff

关 键 词：双能CT 头颈部肿瘤 鳞状细胞癌 淋巴结转移 

分 类 号：R814.42[医药卫生—影像医学与核医学] R739.6[医药卫生—放射医学] R739.8[医药卫生—临床医学]