纹理分析评估肺磨玻璃结节样腺癌的浸润性:基于CT平扫与增强图像的对照研究  

Texture analysis to evaluate the invasiveness of pulmonary ground glass nodular adenocarcinoma:a comparative study of non-enhanced and contrast-enhanced CT imaging

作　　者：吕燕 叶靖[1] 凌俊[1] LU Yan;YE Jing;LING Jun(Department of Radiology,Jiangsu Subei People's Hospital,Jiangsu 225002,China)

机构地区：[1]江苏省苏北人民医院放射科,江苏225002

出　　处：《放射学实践》2021年第12期1503-1508,共6页Radiologic Practice

基　　金：江苏省医学会伦琴影像科研专项资金项目(SYH-3201150-0021-2021016)。

摘　　要：目的:探讨基于CT平扫(NECT)和对比增强(CECT)图像的纹理分析技术对磨玻璃密度(GGO)结节样肺腺癌浸润性的鉴别诊断价值。方法:回顾性分析77例GGO结节样肺腺癌患者的临床和影像资料。77个GGO病灶中,浸润前病变(PIL)12个(15.6%),微浸润性病变(MIA)36个(46.8%),浸润性肺腺癌(IA)29个(37.7%)。分别在每个GGO病灶的CT平扫(NECT)和对比增强(CECT)图像上逐层手动勾画感兴趣区(ROI),获得病灶的容积ROI(VOI),提取其直方图参数(8个)和灰度共生矩阵(GLGM)参数(4个)。采用t检验及受试者工作特征曲线(ROC)分析特征参数组间差异及诊断效能。结果:大多数纹理特征参数对鉴别IA与MIA/PIL具有统计学意义(P<0.05)。ROC曲线分析显示在NECT和CECT纹理参数中,均以能量(AUC分别为0.818和0.839)和熵(AUC分别为0.820和0.859)的诊断效能较高。结论:基于NECT和CECT图像的纹理参数均能较好地鉴别IA与MIA/PIL,其中以能量和熵的诊断效能较好。Objective:To investigate the value of non-enhanced CT(NECT)based and contrast-enhanced CT(CECT)based texture analysis in the differential diagnosis of the invasiveness of lung adenocarcinoma presented as ground-glass opacity(GGO)nodules.Methods:The clinical and CT images of 77 patients with lung adenocarcinoma presented as GGO nodules were retrospectively analyzed.Of the 77 lung GGO nodules,12(15.6%)were pre-invasive lesion(PIL),36(46.8%)were micro-invasive adenocarcinoma(MIA)and 29(37.7%)were invasive adenocarcinoma(IA).The region of interest(ROI)of each GGO was segmented manually layer-by-layer,and then volume ROI(VOI)was obtained,and the texture feature parameters on NECT and CECT images were extracted respectively,including histogram parameters(n=8)and gray level co-occurrence matrix(GLGM)parameters(n=4).T-test and receiver operating characteristic curve(ROC)were used to analyze the difference of the texture features between the PLA and MIA/IA groups,and the diagnostic efficacy of the texture parameters.Results:Whether NECT or CECT texture features,most texture features showed significant differences between IAs and MIAs/PILs(P<0.05).ROC analyses revealed that smaller energy and higher entropy were significant indicators of IA from MIA/PIL,whether based on NECT images[area under the curve(AUC):0.839 and 0.859,respectively]or CECT images(AUC:0.818 and 0.820,respectively).Conclusion:Texture features based on NECT or CECT images have the potential to distinguish IA from PIL/MIA,particularly the parameters of energy and entropy.

关 键 词：磨玻璃结节 体层摄影术 X线计算机 纹理分析 肺腺癌 浸润性 

分 类 号：R814.42[医药卫生—影像医学与核医学] R734.2[医药卫生—放射医学]