增强CT联合纹理分析鉴别无钙化肺错构瘤与原发性肺腺癌的价值  

Enhanced CT Combined with Texture Analysis in Differentiating Non-Calcified Pulmonary Hamartoma from Primary Lung Adenocarcinoma

作　　者：罗成龙 宋一曼 岳松伟[1] 张永高[1] 高剑波[1] 丁昌懋[1] LUO Chenglong;SONG Yiman;YUE Songwei;ZHANG Yonggao;GAO Jianbo;DING Changmao(Department of Radiology,the First Affiliated Hospital of Zhengzhou University,Zhengzhou 450052,China)

机构地区：[1]郑州大学第一附属医院放射科,河南郑州450052

出　　处：《中国医学影像学杂志》2023年第8期838-843,共6页Chinese Journal of Medical Imaging

摘　　要：目的 探讨增强CT联合纹理分析鉴别无钙化肺错构瘤(NCPH)与原发性肺腺癌(PLA)的价值。资料与方法 回顾性收集2018年3月—2021年5月郑州大学第一附属医院经病理证实的NCPH患者84例及PLA患者91例,以7∶3将所有病例随机分为训练集(NCPH 59例,PLA 64例)和验证集(NCPH 25例,PLA 27例)。分析两组CT图像的影像学特征,并采用MaZda软件对CT静脉期薄层图像进行纹理分析。使用筛选出的纹理参数和影像特征构建多因素二元Logistic回归模型,绘制受试者工作特征曲线,计算曲线下面积(AUC),比较各模型对NCPH与PLA的诊断效能。结果 训练集中,CT影像特征模型由平扫CT值、强化程度及毛刺征组成;纳入CT纹理参数模型的最佳纹理参数为Perc.90%和Variance。在训练集中,CT影像特征模型与CT纹理参数模型诊断NCPH与PLA的AUC分别为0.890、0.831,两者差异无统计学意义(P>0.05);影像特征联合纹理参数模型的AUC最大,为0.936,其准确度为0.846。验证集中,CT影像特征模型、CT纹理参数模型及影像特征联合纹理参数模型鉴别两者的AUC分别为0.856、0.834、0.964。结论 CT纹理分析鉴别NCPH与PLA具有一定的诊断价值,联合CT影像特征可进一步提高鉴别效能。Purpose To explore the value of contrast-enhanced CT combined with texture analysis in differentiating non-calcified pulmonary hamartoma(NCPH)from primary lung adenocarcinoma(PLA).Materials and Methods Eighty-four patients with NCPH and 91 patients with PLA confirmed by pathology in the First Affiliated Hospital of Zhengzhou University from March 2018 to May 2021 were collected retrospectively.All patients were randomly divided into a training set(59 cases of NCPH and 64 cases of PLA)and a validation set(25 cases of NCPH and 27 cases of PLA)in a ratio of 7∶3.The imaging features of the two groups of CT images were analyzed,and the MaZda software was used to analyze the texture of CT venous phase thin-layer images.Using the selected texture parameters and image features to construct multi-factor binary Logistic regression models,draw the receiver operating characteristic curve,calculate the area under the curve(AUC),evaluate and compare the diagnostic efficiency of each model for NCPH and PLA.Results In the training set,the CT image feature model was composed of plain scan CT value,enhancement degree and spiculation sign,and the best texture parameters included in the CT texture parameter model were Perc.90%and Variance,respectively.In the training set,the AUC of CT image feature model and CT texture parameter model for diagnosing NCPH and PLA were 0.890 and 0.831,respectively,and there was no significant difference between them(P>0.05).The AUC of image features combined with texture parameter model was the highest,which was 0.936,and its accuracy was 0.846.In the verification set,the AUC of CT image feature model,CT texture parameter model and image features combined with texture parameter model to identify NCPH and PLA were 0.856,0.834 and 0.964,respectively.Conclusion CT texture analysis has a certain diagnostic value in identifying NCPH and PLA,and the combination of CT image features can further improve the identification efficiency.

关 键 词：肺错构瘤 原发性肺腺癌 体层摄影术 X线计算机 纹理分析 诊断 鉴别 

分 类 号：R445.3[医药卫生—影像医学与核医学] R734.2[医药卫生—诊断学]