增强CT纹理分析鉴别食管鳞癌病理分级的应用研究  

Application of enhanced CT-based texture analysis in distinguishing the pathological grade of esophageal squamous cell carcinoma

作　　者：曲长君 杨军乐 董燕[2] 公婷 尹丽娜 党莹 王雨 QU Changjun;YANG Junle;DONG Yan;GONG Ting;YIN Lina;DANG Ying;WANG Yu(Medical College of Xi’an Jiaotong University,Xi’an 710061,China;Department of Radiology,Xi’an Central Hospital,Xi’an Jiaotong University,Xi’an 710003,China;Xi’an First Hospital,Northwest University,Xi’an 710002,China;Shenzhen Anke Company,Shenzhen,Guangdong Province 518000,China;Department of Radiology,Tangdu Hospital,the Air Force Military Medical University,Xi’an 710038,China;Department of Gastroenterology,Xi’an Central Hospital,Xi’an Jiaotong University,Xi’an 710003,China)

机构地区：[1]西安交通大学医学部,陕西西安710061 [2]西安交通大学附属西安市中心医院放射科,陕西西安710003 [3]西北大学附属西安市第一医院,陕西西安710002 [4]深圳安科公司,广东深圳518000 [5]空军军医大学唐都医院放射科,陕西西安710038 [6]西安交通大学附属西安市中心医院消化内科,陕西西安710003

出　　处：《实用放射学杂志》2022年第2期204-208,共5页Journal of Practical Radiology

摘　　要：目的探讨增强CT纹理分析技术鉴别食管鳞癌病理分级的价值.方法回顾性收集62例经手术病理确诊为食管鳞癌患者的临床和增强CT资料,分析其常规影像学特征(包括部位、大小、边界、均匀度及强化程度).利用ITK-SNAP软件在动脉期及静脉期CT图像上画取全肿瘤感兴趣区(ROI),提取食管鳞癌的纹理特征(包括平均值、标准差、能量、偏度、峰度、熵值及自相关).采用单因素分析及多因素Logistic回归对食管鳞癌高、低级别组的临床、常规影像学特征及纹理特征进行比较,并行受试者工作特征(ROC)曲线评估其诊断效能.结果62例食管鳞癌中高级别组23例、低级别组39例.高、低级别食管鳞癌在肿瘤大小、动脉期峰度、熵值、静脉期偏度和自相关的组间比较中存在统计学差异,曲线下面积(AUC)值为0.700~0.803.联合动脉期熵值和静脉期偏度建立的Logistic模型鉴别效能最高,AUC值为0.856.结论基于增强CT的纹理分析有助于鉴别高、低级别食管鳞癌,可为临床病理分级及临床精准治疗提供一定的客观依据.Objective To explore the differential diagnostic value of enhanced CT-based texture analysis in distinguishing the pathological grade of esophageal squamous cell carcinoma.Methods We retrospectively collected the clinical and enhanced CT data of 62 patients with esophageal squamous cell carcinoma confirmed by histopathology,and analyzed the conventional radiologic features,including tumor location,size,boundary,uniformity,and degree of enhancement.Texture features of esophageal squamous cell carcinoma,including mean value,standard deviation,energy,skewness,kurtosis,entropy,and correlation,were extracted from the whole tumor regions of interest(ROI)axial CT images of arterial and venous phase by using ITK-SNAP software.Univariate analysis and multivariate Logistic regression analysis were performed to compare the differences in clinical,conventional radiologic,and texture features between low-grade and high-grade esophageal squamous cell carcinoma.The receiver operating characteristic(ROC)curve was used to evaluate the diagnostic efficacy.Results There were 23 high-grade and 39 low-grade esophageal squamous cell carcinomas in the final cohort.Tumor size,arterial phase kurtosis,entropy,venous phase skewness,and correlation were significantly different for low-grade versus high-grade esophageal squamous cell carcinoma,and the area under the curve(AUC)was 0.700-0.803.The Logistic model based on arterial phase entropy and venous phase skewness performs the highest AUC of 0.856.Conclusion Texture analysis based on enhanced CT is helpful to differentiate low-grade and high-grade esophageal squamous cell carcinoma,and it will provide obj ective evidence for the pathological grade and the clinical precision treatment.

关 键 词：食管鳞癌 计算机体层成像 纹理分析 肿瘤分级 

分 类 号：R735.1[医药卫生—肿瘤] R814.42[医药卫生—临床医学]