摘要
目的探讨CT影像组学术前预测肾透明细胞癌(ccRCC)WHO/ISUP分级的价值。资料与方法回顾性分析330例经病理证实的cc RCC的影像及临床病理资料。根据WHO/ISUP分级将纳入患者分为低级别组(Ⅰ+Ⅱ)及高级别组(Ⅲ+Ⅳ);并按2∶1随机分为训练集和验证集。基于训练集年龄、性别及传统影像特征构建传统影像模型,提取CT髓质期的影像组学特征,在训练集中筛选特征并构建影像组学标签,融合传统影像特征及影像组学标签构建融合模型,并在验证集中验证和评价模型。结果最终提取与ccRCC WHO/ISUP分级相关的2项传统影像特征及27项影像组学特征,分别构建传统影像模型及影像组学标签。ROC曲线下面积显示验证集中影像组学标签优于传统影像模型及融合模型(P<0.05)。校准曲线及Hosmer-Lemeshow检验显示3组模型均具有较好的拟合度(P均>0.05)。决策曲线显示,在验证集中影像组学标签可获得最高的净收益。结论CT影像组学可为ccRCC肿瘤侵袭性的术前评估提供无创、可靠的影像学方法。
Purpose To investigate the value of CT-based radiomics in the preoperative prediction of WHO/ISUP grading of clear cell renal cell carcinoma(ccRCC).Materials and Methods A total of330 patients with ccRCC confirmed by pathology were collected and the imaging and clinicopathological data were retrospectively analyzed.According to the WHO/ISUP grading system,all patients with ccRCC were divided into low-grade group(Ⅰ+Ⅱ)and high-grade group(Ⅲ+Ⅳ).And all patients were randomly divided into training set and validation set at 2:1.A traditional image model was constructed according to the age,gender,and traditional image features in the training set,and the radiomic features of CT imaging in nephrographic phase were extracted.The features in the training set were fiirther selected to construct a radiomic signature.After that,the traditional image features and radiomic signature were combined to build an integrated model.In the end,these models were verified and assessed in the validation set.Results A total of 2 traditional image features and 27 radiomic features associated with ccRCC WHO/ISUP grading were extracted in this study,constructing a traditional imaging model and a radiomic signature,respectively.The receiver operating characteristic(ROC)curve and area under the curve showed that the radiomic signature in the verification set outperformed the traditional imaging model and the integrated model(P<0.05).The calibration curve and Hosmer-Lemeshow test demonstrated that all these three models had a satisfactory degree of fit(P>0.05).The decision curve showed that the radiomic signature had the best outcome in the verification set.Conclusion CT-based radiomics could provide a noninvasive and reliable imaging method for predicting the preoperative aggressiveness in patients with ccRCC.
作者
柏永华
刘衡
张体江
田冲
王荣品
李武超
BO Yonghua;LIU Heng;ZHANG Tijiang;TIAN Chong;WANG Rongpin;LI Wuchao(不详;Department of Radiology,Guizhou Provincial People's Hospital,Guiyang 550002,China;Guizhou Provincial Key Laboratory of Intelligent Medical Image Analysis and Precision Diagnosis,Guiyang 550002,China)
出处
《中国医学影像学杂志》
CSCD
北大核心
2021年第6期585-590,共6页
Chinese Journal of Medical Imaging
基金
国家自然科学基金(81960314)
贵州省人民医院青年基金(GZSYQN[2018]14号)
贵州省卫生健康委员会科学技术基金项目(gzwjkj2019-1-202)。
