基于CT灌注成像影像组学在良恶性孤立性肺结节鉴别诊断中的应用  

Application of radiomics based on CT perfusion imaging in differential diagnosis of benign and malignant solitary pulmonary nodules

作　　者：王德清[1] 赵振华[1] WANG Deqing;ZHAO Zhenhua(Department of Radiology,Shaoring People's Hospital,Shaoring,Zhejiang Province 312000,China)

机构地区：[1]绍兴市人民医院放射科,浙江绍兴312000

出　　处：《实用放射学杂志》2023年第1期37-40,共4页Journal of Practical Radiology

基　　金：绍兴市科技局计划项目(2018C30112)。

摘　　要：目的探讨基于CT灌注成像影像组学在良恶性孤立性肺结节鉴别诊断中的应用价值.方法选取良恶性孤立性肺结节患者82例,根据病理诊断结果分为良性组(40例)和恶性组(42例),比较2组患者CT灌注成像关键参数差异,融合提取的CT灌注成像数据和影像组学特征数据共同建模.结果2组患者的CT灌注参数值比较,恶性组支气管动脉血流量(BF)和肺动脉血流量(PF)高于良性组,良性组灌注指数(PI)高于恶性组,且差异有统计学意义(P<0.05);经过严格筛选、降维处理,共7个特征纳入模型,测试集肺结节联合周围5 mm、10 mm和15 mm模型诊断结果曲线下面积(AUC)值分别为0.88、0.76和0.82,诊断效能皆高于放射科医师,且经Delong检验结果模型AUC均高于低年资医师,差异有统计学意义(P<0.05).结论基于CT灌注成像影像组学可以有效提高良恶性孤立性肺结节的诊断效能,同时将肺结节周围组织纳入建模具有明显优势.Objective To explore the application value of CT perfusion imaging-based radiomics in differential diagnosis of benign and malignant solitary pulmonary nodules.Methods A total of 82 patients with benign and malignant solitay pulmonary nodules were selected and divided into benign group(n=40)and malignant group(n=42)according to pathological diagnosis results.The parameters of CT perfusion imaging between two groups were compared.The models were built by combining the selected CT perfusion parameters and radiomics features datas.Results Compared to the CT perfusion parameters of the two groups,the bronchial flow(BF)and pulmonary flow(PF)of malignant group were higher than those of benign group,and the pulmonary index(PI)of benign group was higher than that of malignant group,and the differences were statistically significant(P<0.05).After rigorous screening and dimensionality reduction,a total of 7 features were included in the model.The area under the curve(AUC)values of diagnosis performances of the combined models of lung nodule and surrounding 5 mm,10 mm or 15 mm tissue in the test set were 0.88,0.76 and O.82,respectively,and the diagnostic efficiency was higher than radiologist,and Delong test results showed that the model AUC was higher than that of junior physicians,and differences were statistically significant(P<0.05).Conclusion CT perfusion imaging-based radiomics can effectively improve the diagnostic efficiency of benign and malignant solitary pulmonary nodules.And the radiomics features of combing lung nodule and surrounding tissues can improve the model performance.

关 键 词：CT灌注成像 影像组学 肺结节 

分 类 号：R814.42[医药卫生—影像医学与核医学] R445[医药卫生—放射医学] R563[医药卫生—临床医学]