基于CT影像组学鉴别局灶性隐源性机化性肺炎与肺腺癌  

Differential diagnosis of focal cryptogenic organized pneumonia and lung adenocarcinoma based on CT radiomics

作　　者：王丽丽 郑祥武[2] 唐坤[2] 陈雀芦 WANG Lili;ZHENG Xiangwu;TANG Kun(Department of Radiology,Wenzhou Central Hospital,Wenzhou 325000,China)

机构地区：[1]温州市中心医院放射科,浙江温州325000 [2]温州医科大学附属第一医院放射科

出　　处：《全科医学临床与教育》2024年第1期15-19,F0003,共6页Clinical Education of General Practice

基　　金：温州市基础性科研项目(Y20220461)。

摘　　要：目的 基于影像组学建立列线图鉴别局灶性隐源性机化性肺炎(FCOP)与肺腺癌(LA)。方法 回顾性搜集77例FCOP与77例LA的影像及临床资料,按照7∶3随机分为训练集106例和验证集48例。采用A.K.软件提取组学特征,降维后获得影像组学标签,建立列线图,并评估模型的诊断效能、校准度及临床应用价值。结果 11个高度相关的组学特征构建影像组学标签,并联合癌胚抗原(CEA)建立列线图,训练集和验证集的受试者工作特征曲线(ROC)下面积均为0.84,且具有较高的校准度及临床应用价值。结论 基于CT影像组学特征建立的列线图能较好地用于鉴别FCOP与LA。Objective To establish a nomogram based on CT radiomics in differential diagnosis of focal cryptogenic organizing pneumonia(FCOP) and lung adenocarcinoma(LA).Methods The CT and clinical data of 77 patients with FCOP and 77 patients with LA were collected retrospectively.The patients were randomly divided into training set(n=106) and verification set(n=48) according to 7:3.Radiomics features are extracted by A.K.software and constructed radiomics signature after the screening.Then we constructed the nomogram,and evaluated the diagnostic efficiency,internal validation,calibration assessment and clinical usefulness of the model.Results The eleven highly correlated radiomics features were screening to construct radimomics signature.The AUC of the nomogram of CEA and radiomics signature in the training group and the verification group was 0.84.And the nomogram showed good diagnostic efficiency,internal validation and clinical usefulness of the model.Conclusion The nomogram established by CT radiomics features has good performance to distinguish FCOP and LA.

关 键 词：局灶性隐源性机化性肺炎 肺腺癌 影像组学 列线图 

分 类 号：R734.2[医药卫生—肿瘤] R730.44[医药卫生—临床医学]