CT影像组学联合临床—影像特征的列线图鉴别腺性膀胱炎与膀胱癌  

Differentiation on cystitis glandularis and bladder cancer based on nomogram of CT imaging radiomics combined with clinical-imaging features

作　　者：刘雪成 吴树剑 姚琪 王娟 Liu Xuecheng;Wu Shujian;Yao Qi;Wang Juan(Department of Radiology,the First Affiliated Hospital of Wannan Medical College,Wuhu,Anhui 241000,China)

机构地区：[1]皖南医学院第一附属医院放射科,安徽芜湖241000

出　　处：《齐齐哈尔医学院学报》2025年第3期267-273,共7页Journal of Qiqihar Medical University

摘　　要：目的 探讨基于临床—影像特征联合增强CT影像组学列线图鉴别腺性膀胱炎与膀胱癌的价值。方法 回顾性分析2012年4月—2023年10月本院收治的经病理证实的291例腺性膀胱炎(91例)与膀胱癌(200例)患者的临床及影像学资料,将患者按7︰3的比例随机分为训练集(n=204)和验证集(n=87)。使用开源软件FAE从增强CT静脉期图像中提取组学特征,通过降维筛选最优特征建立组学标签,并计算标签得分(Radscore);行单因素和多因素Logistic回归分析用于筛选独立预测因素,基于独立预测因素分别构建临床模型、影像组学模型和临床—组学联合模型,并绘制联合模型列线图。使用受试者工作特征曲线下面积评估模型的诊断效能;决策曲线用于评价模型临床净收益;DeLong检验用于比较各模型AUC差异;使用校正曲线评估模型的拟合度。结果 经特征降维筛选出7个最优组学特征。回归分析显示年龄、强化程度和Radscore为独立预测因素。在训练集中,联合模型列线图的AUC为0.895,高于影像组学模型(0.863)和临床模型(0.797);且DeLong检验比较差异均有统计学意义(Z=1.983,P=0.047;Z=3.455,P<0.001)。联合模型在训练组(P=0.326)和验证组(P=0.419)均拟合良好。联合模型的临床净收益均高于临床模型和影像组学模型。结论 基于临床—影像特征联合增强CT组学的列线图能够有效区分腺性膀胱炎与膀胱癌。Objective To explore the value of clinical-imaging features combined with enhanced CT imaging radiomics in differentiating cystitis glandularis from bladder cancer.Methods Retrospective analysis was conducted on the clinical and imaging data of 291 patients with pathologically confirmed cystitis glandularis(91 cases)and bladder cancer(200 cases)at the First Affiliated Hospital of Wannan Medical College from April 2012 to October 2023.Patients were randomly divided into a training set(n=204)and a validation set(n=87)with a ratio of 7:3.Open-source software FAE was used to extract radiomics features from enhanced CT venous phase images,and optimal features were selected through dimensionality reduction to establish a radiomics label and calculate the label score(Radscore).Single-factor and multi-factor logistic regression analyses were conducted to identify independent predictive factors.Clinical,radiomics,and combined clinical-radiomics models were constructed based on these independent predictive factors,along with a nomogram of the combined model.The diagnostic performance of the models was assessed using the area under the receiver operating characteristic curve(AUC);decision curves were used to evaluate the clinical net benefit of the models;DeLong test was employed to compare differences in AUC among models;and calibration curves were used to assess the goodness-of-fit of the models.Results Seven optimal radiomics features were identified through dimensionality reduction.Regression analysis indicated that age,enhancement degree,and Radscore were independent predictors.In the training set,the AUC of the combined model nomogram was 0.895,higher than that of the radiomics model(0.863)and the clinical model(0.797).The differences were statistically significant according to DeLong's test(Z=1.983,P=0.047;Z=3.455,P<0.001).The combined model exhibited good fit in both the training group(P=0.326)and validation group(P=0.419).The clinical net benefits of the combined model were higher than those of the clinical model a

关 键 词：腺性膀胱炎 膀胱癌 影像组学 列线图 计算机体层摄影技术 

分 类 号：R73[医药卫生—肿瘤]