CT影像组学在诊断囊腔型肺腺癌浸润程度及建模的价值

Value of radiomics in diagnosing the invasion degree and modeling of cystic lung adenocarcinoma

目的探讨CT影像组学在诊断囊腔型肺腺癌浸润程度及建模的价值,以加深对该类肺癌的认识,指导临床进行分层管理。方法选取经病理确诊的32例囊腔型肺腺癌患者,依据病理结果分为相对良性组(14例)和浸润组(18例)。利用3D Slicer软件沿病灶边缘手动勾画感兴趣区(region of interest,ROI),利用pyradiomics软件对获得的ROI进行特征提取,采用最小绝对收缩和选择算子(least absolute shrinkage and selection operator,LASSO)回归、t检验及向后逐步回归(backward stepwise)进行特征降维、筛选,以P<0.05为差异有统计学意义。对于筛选出的预测因子,用SPSS25.0计算ROC曲线下面积(area underthe curve,AUC),以评价列线图鉴别囊腔型肺腺癌的效能,并采用R软件的survival及rms工具包绘制列线图,同时采用Bootstrap法对列线图模型的预测性能进行检验。结果最短短轴(Shape_LeastAxisLength)、最长短轴(Shape_MinorAxisLength)两个影像组学特征能较好预测囊腔型肺腺癌的浸润程度。结论CT影像组学特征对预测囊腔型肺腺癌的浸润程度具有重要诊断价值,有望为临床提供一种无创、量化、方便快捷的诊断方法。

Objective To explore the value of radiomics in diagnosing the invasion degree and modeling of cystic lung ad⁃enocarcinoma,to deepen the understanding of this type of lung cancer,and to guide clinical stratified management.Methods 32 cases of cystic lung adenocarcinoma confirmed by pathology were retrospectively analyzed.According to pathological results,cystic lung adenocarcinoma was divided into relatively benign group(14 cases)and invasive group(18 cases).3D Slicer soft⁃ware was used to manually delineate the region of Interest(ROI)along the edge of the lesion,and Pyradiomics software was used to extract the ROI obtained.The least absolute shrinkage and Selection operator(LASSO)regression,T test and backward stepwise were used for feature dimensionality reduction and screening.P<0.05 was considered statistically significant.For the se⁃lected predictors,SPSS25.0 was used to calculate the area under the ROC curve(AUC)to evaluate the effectiveness of the no⁃mograph in differentiating cystic lung adenocarcinoma,and the survival and RMS toolkit of R software were used to draw the no⁃mograph.At the same time,Bootstrap method was used to test the prediction performance of the line graph model.Results Shape_LeastAxisLength and Shape_MinorAxisLength could better predict the invasion degree of cystic lung adenocarcinoma.Conclusion CT imaging features have important diagnostic value in predicting the invasion degree of cystic lung adenocarci⁃noma,which is expected to provide a non-invasive,quantitative,convenient and rapid diagnostic method for clinic.