陆亮,徐圆,袁为标,陆大军.增强CT影像组学列线图预测膀胱尿路上皮癌肌层浸润[J].中国医学影像技术,2021,37(7):1059~1063
增强CT影像组学列线图预测膀胱尿路上皮癌肌层浸润
Enhanced CT radiomics nomogram for predicting muscular infiltration of bladder urothelial carcinoma
投稿时间:2020-07-27  修订日期:2021-04-18
DOI:10.13929/j.issn.1003-3289.2021.07.022
中文关键词:  膀胱肿瘤  癌,移行细胞  体层摄影术,X线计算机  影像组学
英文关键词:bladder neoplasm  carcinoma, transitional cell  tomography, X-ray computed  radiomics
基金项目:
作者单位E-mail
陆亮 南通大学附属建湖人民医院影像科, 江苏 盐城 224700 xy86463106@sina.com 
徐圆 盐城市第一人民医院影像科, 江苏 盐城 224006  
袁为标 南通大学附属建湖人民医院影像科, 江苏 盐城 224700  
陆大军 南通大学附属建湖人民医院影像科, 江苏 盐城 224700  
摘要点击次数: 228
全文下载次数: 240
中文摘要:
      目的 评价增强CT影像组学列线图预测膀胱尿路上皮癌肌层浸润的可行性。方法 纳入91例膀胱尿路上皮癌患者,根据手术病理结果分为非肌层浸润性膀胱癌(NMIBC)组(n=51)和肌层浸润性膀胱癌(MIBC)组(n=40),比较组间CT特征差异。利用Mazda软件提取病变纹理参数,以Lasso算法筛选,联合十折交叉验证构建Logistic回归影像组学列线图。绘制受试者工作特征(ROC)曲线,计算曲线下面积(AUC),评估列线图预测膀胱尿路上皮癌肌层浸润的效能。结果 CT形态不规则多见于MIBC组,NMIBC组与MIBC组肿瘤CT形态特征差异具有统计学意义(P<0.05)。Logistic回归组学列线图预测肌层浸润AUC为0.881,特异度为76.5%,敏感度87.5%,危险因素包括动脉期S(3,3)SumAverg、S(4,-4)InvDfMom及静脉期S(3,-3)DifEntrp、Perc.90%。结论 增强CT影像组学列线图有助于术前预测膀胱尿路上皮癌肌层浸润。
英文摘要:
      Objective To evaluate the feasibility of enhanced CT radiomics nomogram for predicting muscular infiltration of bladder urothelial carcinoma. Methods A total of 91 patients with bladder urothelial carcinoma were enrolled. According to postoperation pathological results, the patients were divided into non-muscle invasive bladder cancer (NMIBC) group (n=51) and muscle invasive bladder cancer (MIBC) group (n=40), and CT features were compared between groups. Texture parameters were extracted with Mazda software, while feature parameters were extracted with Lasso algorithm. The Logistic regression radiomics nomogram was constructed combining with ten-fold cross validation. The receiver operating characteristic (ROC) curves were drawn, and the area under the curves (AUC) was calculated to evaluate the efficacy of the nomogram for predicting muscular infiltration of bladder urothelial carcinoma. Results Among CT features, irregular morphology was more common in MIBC group than in NMIBC group. Significant differences of morphological characteristics were found between groups (P<0.05). The AUC, specificity and sensitivity of Logistic regression radiomics nomogram for predicting muscular infiltration was 0.881, 76.5% and 87.5%, respectively. The risk factors included S(3,3)SumAverg and S(4,-4)InvDfMom in arterial phase, and S(3,-3)DifEntrp and Perc.90% in venous phase. Conclusion Enhanced CT radiomics nomogram was helpful for preoperative prediction of muscular infiltration of bladder urothelial carcinoma.
查看全文  查看/发表评论  下载PDF阅读器