摘要
目的本研究通过CT形态学特征、CT纹理分析探讨其对肺良、恶性磨玻璃结节(GGN)的诊断价值及诊断效能。方法采用回顾性分析方法,搜集180例经本院手术切除并取得病理结果的肺部GGN患者。根据病理结果分为两组(良性组76例,恶性组104例)。将患者术前CT图像数据分别导入技术软件得到肺结节形态学及CT纹理分析参数结果,比较这两种分析参数的定性和定量结果,并基于肺结节形态学、CT纹理分析参数建立二元Logistic回归模型,采用受试者工作特征曲线(ROC)比较两种模型以及联合模型对良、恶性肺结节的诊断效能。结果肺结节形态学显示长径(cm)、短径(cm)、平均径(cm)、体积(mm^(3))、实性部分占位(%)、最大CT值(HU)、最小CT值(HU)、平均CT值(HU)、分叶征、毛刺征、胸膜凹陷征,良、恶性组间比较有统计学差异(P<0.05);经CT纹理分析及组间比较共合计获得36个有意义参数(P<0.05);二元Logistic回归分析显示:肺结节形态学的分叶征、毛刺征,CT纹理分析的正像素平均值(MPP)、容积计数(Volume Count)、相关性(correlation)、表面积(SurfaceArea)(P<0.05),有助于肺部GGN良、恶性的诊断;肺结节形态学与CT纹理分析的联合模型,较肺结节形态学模型、CT纹理分析模型具有更高的诊断效能[曲线下面积(AUC)0.911 vs.0.824,P<0.05;AUC 0.911 vs.0.830,P<0.05]。结论肺结节形态学、CT纹理分析对肺良恶性磨玻璃结节的鉴别诊断具有一定价值,其联合模型具有更高的诊断效能。
Objective The present study aimed to evaluate the diagnostic value of CT morphologic feature combined with CT texture analysis for benign and malignant ground glass nodules(GGNs).Methods We retrospectively reviewed 129 patents who were pathologically diagnosed.According to the results of pathologic diagnosis,all patents were divided into two groups:malignant(n=32)and benign(n=97).Preoperative computed tomography images were imported into imaging processing software for morphology analysis and CT texture analysis.The results were compared qualitatively and quantitatively.Binary Logistic regression models were applied to analysis the data.The Receiver Operating Characteristic curve was used to assess the predictive ability of the models.Results Among CT morphology analysis features,long diameter(cm),short diameter(cm),average diameter(cm),volume(mm~3),solid part occupation(%),maximum CT value(HU),minimum CT value(HU),average CT value(HU),lobular sign,burr sign,pleural depression sign,there were statistically significant differences between benign and malignant groups(P<0.05).A total of 36 CT texture features showed significant difference by intergroup compared(P<0.05).Binary Logistic regression analysis showed that leaf sign、glitch sign、MPP、Volume Count、Correlation、SurfaceArea were helpful to the diagnosis of benign and malignant GGNs.The combined model of pulmonary nodule morphology and CT texture analysis had higher diagnostic efficiency than pulmonary nodule morphology model and CT texture analysis model(Area under the curve,AUC=0.911 vs.AUC=0.829,P<0.05;AUC=0.911 vs.AUC=0.830,P<0.05).Conclusion Pulmonary nodule morphology and CT texture analysis have certain clinical diagnostic value in pulmonary benign and malignant ground glass nodules,and the combined model has higher diagnostic efficiency.
作者
臧汉杰
郭浩东
王之悦
朱建国
李海歌
刘剑
唐继来
ZANG Hanjie;GUO Haodong;WANG Zhiyue(Department of Medical Imaging,The Second Affiliated Hospital of Nanjing Medical University,Nanjing,Jiangsu Province 210011,P.R.China)
出处
《临床放射学杂志》
北大核心
2022年第4期628-634,共7页
Journal of Clinical Radiology
关键词
肺腺癌
磨玻璃结节
形态学
体层摄影术
X线计算机
纹理分析
Lung adenocarcinoma
Ground glass nodules
Imaging features
Tomography,X-ray computed
Texture analysis
作者简介
通讯作者:刘剑;通讯作者:唐继来。
