双层探测器光谱CT平扫多参数诊断急性肺栓塞的价值  

The value of non-contrast dual-layer spectral detector CT with multiple parameters in the diagnosis of acute pulmonary embolism

作　　者：谢斯栋[1] 储圆圆 段亚妮 朱雁秋 康庄[1] 张建生[1] 覃杰[1] Xie Sidong;Chu Yuanyuan;Duan Yani;Zhu Yanqiu;Kang Zhuang;Zhang Jiansheng;Qin Jie(Department of Radiology,the Third Affiliated Hospital of Sun Yat-Sen University,Guangzhou 510630,China)

机构地区：[1]中山大学附属第三医院放射科,广州510630

出　　处：《中华放射学杂志》2023年第10期1080-1086,共7页Chinese Journal of Radiology

基　　金：广东省自然科学基金(2017A030313841);中山大学附属第三医院国自然基金培育项目(2021GZRPYM06);中山大学附属第三医院“五个五”工程项目(2023WW605)。

摘　　要：目的探讨双层探测器光谱CT平扫虚拟单能量图像(VMI)及电子云密度图(EDM)在急性肺栓塞诊断中的价值。方法回顾性分析2022年10月至2023年5月在中山大学附属第三医院接受双层探测器光谱CT肺动脉血管造影(CTPA)且证实为急性肺栓塞的27例患者(共计41个病灶)的临床和影像学资料。所有患者均接受光谱CT平扫,并基于平扫图像分别重建常规120 kVp混合能量图像(PI)、虚拟单能量图(40、70、100 keV VMI)、电子云密度图(EDM)和有效原子序数图(Z_(eff))。以CTPA为金标准,评估不同重建图像上肺栓塞的检出率。以血管病变为肺栓塞组,以血管病变同级别正常血管相对应位置为对照组,采用Wilcoxon秩和检验比较2组PI及40、70、100 keV VMI上CT值(CT_(PI)、CT_(40 keV)、CT_(70 keV)、CT_(100 keV))及EDM上ED值、Z_(eff)图上Z_(eff)值的差异。将差异有统计学意义的参数纳入多因素logistic回归,最终构建logistic回归模型。采用受试者操作特征(ROC)曲线和曲线下面积(AUC)分析不同定量参数及logistic回归模型鉴别诊断肺栓塞组与对照组的效能。结果在PI、EDM、Z_(eff)图、EDM与Z_(eff)图融合图像上肺栓塞的检出率分别为14.6%(6/41)、82.9%(34/41)、51.2%(21/41)、97.6%(40/41)。肺栓塞组的CT_(PI)、ED、CT_(40 keV)、CT_(70 keV)和CT_(100 keV)均高于对照组(Z值分别为1009.00、1024.50、1038.00、1079.00、1076.00,P<0.05)。最终筛选出CT_(PI)、CT_(40 keV)和CT_(100 keV) 3个特征构建logistic回归模型,其鉴别栓塞组与对照组的AUC为0.771,灵敏度为0.769,特异度为0.744,准确度为0.756。结论双层探测器光谱CT平扫VMI及EDM对检出和诊断急性肺栓塞具有一定的临床价值。Objective To explore the diagnostic value of virtual monoenergetic image(VMI)and electron density map(EDM)generated by non-contrast dual-layer spectral detector CT in acute pulmonary embolism(APE).Methods The clinical and imaging data of 27 patients(41 lesions)who underwent CT pulmonary angiography(CTPA)using dual-layer spectral detector CT and were diagnosed with APE in the Third Affiliated Hospital of Sun Yat-sen University from October 2022 to May 2023 were retrospectively analyzed.All patients received a dual-layer spectral detector CT non-contrast scan.Based on the non-contrast scan data,conventional 120 kVp polyenergetic images(PI),virtual monoenergetic images(40,70,100 keV VMI),electron density maps(EDM),and effective atomic number maps(Z_(eff))were respectively reconstructed.Taking CTPA as the gold standard,the detection rate of APE in different reconstruction images of non-contrast scan were evaluated.The vascular lesion was used as the pulmonary embolism group and the corresponding position of the normal vessel at the same level of the diseased vessel as the control group,and the Wilcoxon rank-sum test was used to compare the differences of CT values on PI and 40,70,and 100 keV VMI(CT_(PI),CT_(40 keV),CT_(70 keV),CT_(100 keV))as well as the ED values on the EDM,and the Z_(eff) value on the Z_(eff) images between the 2 groups.Parameters with statistically significant differences were included in a multifactor logistic regression,resulting in the construction of a logistic regression model.Receiver operator characteristic curve and area under curve(AUC)were applied to evaluate the diagnostic efficiency of different spectral quantitative parameters and logistic regression model in identifying pulmonary embolism group from normal control group.Results The detection rates of APE on PI,EDM,Z_(eff) map,and EDM and Z_(eff) map fusion images were 14.6%(6/41),82.9%(34/41),51.2%(21/41),and 97.6%(40/41),respectively.CT_(PI),ED,CT_(40 keV),CT_(70 keV) and CT_(100 keV) in pulmonary embolism group were statistically h

关 键 词：肺栓塞 体层摄影术 X线计算机 虚拟单能量成像 电子云密度 

分 类 号：R563.5[医药卫生—呼吸系统]