基于ADC直方图特征的列线图模型在预测移行区临床显著性前列腺癌中的应用  

Application of nomogram model based on ADC histogram features in predicting clinically significant prostate cancer in transitional zone

作　　者：章双林 陈昉铭 高茜 ZHANG Shuanglin;CHEN Fangming;GAO Xi(Department of Radiology,Jiangnan University Medical Center,Wuxi 214002,China)

机构地区：[1]江南大学附属中心医院影像科,无锡214002

出　　处：《磁共振成像》2025年第4期87-92,共6页Chinese Journal of Magnetic Resonance Imaging

摘　　要：目的 用表观扩散系数(apparent diffusion coefficient,ADC)直方图特征开发一种列线图模型,以预测移行区临床显著性前列腺癌(clinically significant prostate cancer,CSPCa)。材料与方法 回顾性分析我院2019年1月至2024年6月泌尿外科收治的283例可疑前列腺癌患者的临床及影像资料。患者被随机分为训练集(70%,198例)和内部验证集(30%,85例)。应用最小绝对收缩和选择算子(least absolute shrinkage and selection operator,LASSO)算法筛选出关键特征:ADC_最小值(apparent diffusion coefficient minimum,ADC_min)、ADC变异系数(coefficient of variation of apparent diffusion coefficient,ADC_CoeffOfVar)、ADC_峰度(apparent diffusion coefficient kurtosis,ADC_kurtosis)、ADC_熵(apparent diffusion coefficient entropy,ADC_entropy),并进一步应用单因素和多因素logistic回归分析筛选变量,构建预测模型。以受试者工作特征(receiver operating characteristic,ROC)曲线下面积(area under the curve,AUC)、敏感度、特异度、阳性预测值、阴性预测值、准确度评价诊断效能,并通过决策曲线分析(decision curve analysis,DCA)评估临床净效益。结果 研究发现,ADC_CoeffOfVar [比值比(odds ratio,OR)=1.01,P=0.034]和ADC_entropy (OR=1.00,P<0.001)是CSPCa的独立预测因子。基于这些因子构建的列线图模型在训练集(AUC=0.844)和内部验证集(AUC=0.765)中均展现出良好的预测性能。校准曲线分析表明模型预测与实际观察结果高度一致,DCA进一步证实了模型在临床决策中的净效益。结论 基于ADC直方图特征构建的列线图模型不仅为术前风险评估提供了一种无创工具,而且具有实际的临床应用潜力。Objective:To develop a nomogram model using apparent diffusion coefficient(ADC) histogram features to predict clinically significant prostate cancer(CSPCa) in the transition zone.Materials and Methods:A retrospective analysis was conducted on 283patients with suspicious prostate cancer admitted to the urology department of our hospital from January 2019 to June 2024.The patients were randomly divided into a development set(70%,198 cases) and an internal validation set(30%,85 cases).The least absolute shrinkage and selection operator(LASSO) algorithm was applied to screen for key features:ADC_min(apparent diffusion coefficient minimum),ADC_CoeffOfVar(coefficient of variation of apparent diffusion coefficient),ADC_kurtosis(apparent diffusion coefficient kurtosis) and ADC_entropy(apparent diffusion coefficient entropy).Furthermore,univariate and multivariate logistic regression analyses were performed to select variables and construct a predictive model.Diagnostic performance was evaluated using area under the curve(AUC) of the receiver operating characteristic(ROC),sensitivity,specificity,positive predictive value,negative predictive value,and accuracy.Decision curve analysis(DCA) was also employed to assess clinical net benefit.Results:ADC_CoeffOfVar [odds ratio(OR) =1.01,P = 0.034] and ADC_entropy(OR = 1.00,P < 0.001) were independent predictors of CSPCa.The nomogram model constructed based on these factors demonstrated good predictive performance in both the development set(AUC = 0.844) and the internal validation set(AUC = 0.765).Calibration curve analysis showed a high degree of agreement between model predictions and actual observations,and decision curve analysis further confirmed the net benefit of the model in clinical decision-making.Conclusions:The nomogram model constructed based on ADC histogram features not only provides a non-invasive tool for preoperative risk assessment but also holds practical clinical application potential.

关 键 词：前列腺肿瘤 临床显著性前列腺癌 磁共振成像 列线图 

分 类 号：R445.2[医药卫生—影像医学与核医学] R737.25[医药卫生—诊断学]