能谱CT多参数定量分析预测原发性肺癌病理类型

Multi-parameter quantitative analysis of spectral CT for predicting pathological types of primary lung cancer

目的观察能谱CT多参数定量分析预测原发性肺癌病理类型的价值。方法纳入137例接受能谱CT检查的肺癌患者,根据病理结果分为腺癌组(n=74)、鳞癌组(n=39)及小细胞肺癌(SCLC)组(n=24)。对比各组能谱CT参数,包括病灶40~100 keV间隔为10 keV的单能量CT值、碘浓度(IC)、标准化碘浓度(NIC)及能谱曲线斜率等的差异。绘制受试者工作特征(ROC)曲线,计算曲线下面积(AUC),分析各参数预测肺癌病理类型的效能。结果动脉期3组间40 keV单能量CT值、IC、NIC及能谱曲线斜率差异均有统计学意义(P均<0.05);SCLC组40 keV单能量CT值、IC、NIC及能谱曲线斜率与腺癌组、鳞癌组差异均具有统计学意义(P均<0.05)。静脉期3组间40~100 keV每间隔10 keV单能量CT值、IC、NIC及能谱曲线斜率差异均有统计学意义(P均<0.05);SCLC组40~90 keV每间隔10 keV的单能量CT值、IC、NIC及能谱曲线斜率与腺癌组、鳞癌组差异均具有统计学意义(P均<0.05)。据此将腺癌组和鳞癌组合并为非小细胞肺癌(NSCLC)组,根据上述差异具有统计学意义的参数诊断SCLC与NSCLC的AUC均>0.7;联合动、静脉期40 keV单能量CT值、NIC及能谱曲线斜率诊断SCLC与NSCLC的AUC分别为0.809和0.855。结论能谱CT多参数联合定量分析预测原发性肺癌的病理类型具有较高效能。

Objective To observe the value of multi-parameter quantitative analysis of spectral CT for predicting the pathological types of primary lung cancer.Methods A total of 137 patients with lung cancer who underwent energy spectral CT examination were enrolled and divided into adenocarcinoma group(n=74),squamous cell carcinoma group(n=39)and small cell lung cancer(SCLC)group(n=24)according to pathological results.The energy spectral CT parameters,including 40-100 keV monochromatic CT value of interval of 10 keV,iodine concentration(IC),normalized iodine concentration(NIC)and slope of spectral attenuation curves,etc.were compared among groups.Receiver operating characteristic(ROC)curves were drawn,and the area under the curve(AUC)were calculated to analyze the efficacy of each parameter in predicting the pathological types of lung cancer.Results In arterial phase,there were statistically significant differences of 40 keV monochromatic CT value,IC,NIC and slope of energy spectrum curve among 3 groups(all P<0.05),also of 40 keV monochromatic CT value,IC,NIC and slope of spectral attenuation curves between SCLC group and adenocarcinoma group as well as between SCLC group and squamous cell carcinoma group(all P<0.05).In venous phase,there were statistically significant differences of 40-100 keV monochromatic CT value,IC,NIC and slope of spectral attenuation curves among 3 groups(all P<0.05),also of 40-90 keV monochromatic CT value,IC,NIC and slope of spectral attenuation curves between SCLC group and adenocarcinoma group,as well as between SCLC group and squamous cell carcinoma group(all P<0.05).Adenocarcinoma group and squamous cell carcinoma group were thereafter combined into non-small cell lung cancer(NSCLC)group.The AUCs of above parameters being statistical different were all greater than 0.7 in differentiating SCLC and NSCLC.Combining 40 keV monochromatic CT value,NIC and slope of spectral attenuation curves in arterial and venous phase,the AUC of predicting SCLC and NSCLC was 0.809 and 0.855,respectively.Conclusion Multi-parameter combining quantitative analysis of spectral CT had high efficacy in predicting pathological types of primary lung cancer.