CT纹理特征参数对不同浸润程度亚实性结节肺腺癌的诊断价值

Diagnostic value of CT texture feature parameters in lung adenocarcinoma of different degree of infiltration with subsolid nodules

目的:分析CT纹理特征参数对不同浸润程度亚实性结节(SNN)肺腺癌的诊断价值。方法:回顾性分析2017年1月至2021年12月该院收治的83例SNN肺腺癌患者的临床资料,均接受CT检查和病理检查,提取CT纹理特征参数。根据病理结果,统计不同浸润程度SNN肺腺癌患者例数,并分为微浸润性腺癌(MIA)组和浸润性腺癌(IAC)组,比较两组一阶纹理特征参数(峰度、偏度、标准差)、二阶灰度共生矩阵参数(熵、逆差距、对比度、自相关)、容积纹理特征参数(质量、表面积、最大有效长径);绘制受试者工作特征(ROC)曲线,分析CT纹理特征参数对不同浸润程度SNN肺腺癌的诊断价值。结果:病理检查结果显示,MIA 40例,IAC 43例;两组峰度、偏度、对比度、自相关值、表面积、最大有效长径比较,差异均无统计学意义(P>0.05);IAC组标准差、熵值、质量值均大于MIA组,逆差距值小于MIA组,差异有统计学意义(P<0.05);ROC曲线结果显示,逆差距、质量、标准差、熵单项及联合应用诊断不同浸润程度SNN肺腺癌患者的AUC为0.882、0.835、0.777、0.840、0.995,联合应用诊断价值最高。结论:CT纹理特征中质量、逆差距、熵、标准差参数对不同浸润程度SNN肺腺癌患者均具有一定的诊断价值,且联合应用诊断价值更高。

Objective:To analyze diagnostic value of CT texture feature parameters in lung adenocarcinoma of different degrees of infiltration with subsolid nodule(SNN).Methods:The clinical data of 83 patients with lung adenocarcinoma and SNN admitted to the hospital from January 2017 to December 2021 were retrospectively analyzed.All patients underwent CT examination and pathological examination to extract the CT texture feature parameters.According to the pathological results,the number of patients with lung adenocarcinoma of different degrees of infiltration and SNN was statistically analyzed and were divided into microinvasive adenocarcinoma(MIA)and invasive adenocarcinoma(IAC)groups.The first order texture feature parameters(kurtosis,skewness,standard bias),the second order gray level cooccurrence matrix parameters(entropy,inverse different moment,contrast,autocorrelation),and the volume texture feature parameters(mass,surface area,maximum effective length)were compared between the two groups.The receiver operating characteristic curve(ROC)was drawn to analyze the diagnostic value of CT texture feature parameters for SNN lung adenocarcinoma of different degrees of infiltration with SNN.Results:The pathological examination showed 40 cases of microinvasive adenocarcinoma(MIA)and 43 cases of invasive adenocarcinoma(IAC).There were no significant differences in kurtosis,skewness,contrast,autocorrelation value,surface area and maximum effective length between the two groups(P>0.05).The standard deviation,entropy value and mass value of the IAC group were higher than those of the MIA group,and the inverse different moment value was lower than that of the MIA group,and the differences were statistically significant(P<0.05).The ROC results showed that the AUCs of inverse different moment,mass,standard deviation,entropy and combined application in the diagnosis of lung adenocarcinoma of different degrees of infiltration with SNN were 0.882,0.835,0.777,0.840 and 0.995,respectively,and the combined application had the highest diagnostic value.Conclusions:The mass,inverse different moment,entropy and standard deviation parameters of CT texture features have certain diagnostic value for lung adenocarcinoma of different degrees of infiltration with SNN,and the combined application has higher diagnostic value.