基于多模态超声参数的Logistic回归模型诊断前列腺癌的临床价值  

Clinical value of Logistic regression model based on multimodal ultrasound parameters in the diagnosis of prostate cancer

作　　者：周隽 郑兵[2] ZHOU Jun;ZHENG Bing(Department of Ultrasound,the Second Affiliated Hospital of Nantong University,Nantong 226000,China;Department of Urology,the Second Affiliated Hospital of Nantong University,Nantong 226000,China)

机构地区：[1]南通大学第二附属医院超声科,江苏南通226000 [2]南通大学第二附属医院泌尿外科,江苏南通226000

出　　处：《临床超声医学杂志》2025年第3期232-237,共6页Journal of Clinical Ultrasound in Medicine

摘　　要：目的 基于多模态超声参数构建诊断前列腺癌(PCa)的Logistic回归模型,探讨其临床应用价值。方法 选取我院收治的疑似PCa患者180例(训练集),依据病理结果分为PCa组93例和非PCa组87例,应用二维超声观察病灶形态、内部回声、包膜、结节、钙化,应用CDFI对血流分级情况,计算二维超声和CDFI图像阳性率;剪切波弹性成像获取病灶杨氏模量最大值、平均值、最小值(Emax、Emean、Emin);超声造影获取造影剂到达时间、达峰时间、基础强度、峰值强度、强度差、增强强度,比较两组上述检查结果和临床资料的差异。采用多因素Logistic回归分析筛选诊断PCa的独立影响因素,并构建诊断PCa的Logistic回归模型。绘制受试者工作特征(ROC)曲线分析模型的诊断效能;采用HosmerLemeshow拟合优度检验评估模型的拟合度。另选32例疑似PCa患者作为验证集对模型进行外部验证。结果 PCa组血清前列腺特异性抗原、前列腺特异性抗原密度(PSAD)、二维超声和CDFI图像阳性率、Emax、Emean、Emin,以及超声造影峰值强度、强度差、增强强度均高于非PCa组,差异均有统计学意义(均P<0.05);两组其余参数比较差异均无统计学意义。多因素Logistic回归分析显示,血清PSAD、Emax和强度差均为诊断PCa的独立影响因素(均P<0.05),构建的Logistic回归模型为:Logit(P)=1/[1+e^((-10.369+1.134×血清PSAD+×1.359×Emax+1.089×强度差))]。ROC曲线分析显示,模型诊断训练集中PCa的曲线下面积为0.951(95%可信区间:0.929~0.974,P<0.05);Hosmer-Lemeshow拟合优度检验显示模型具有较好的拟合度(χ^(2)=3.696,P=0.883)。外部验证结果显示,模型诊断验证集中PCa的曲线下面积为0.932(95%可信区间:0.876~0.988,P<0.05)。结论 基于多模态超声参数构建的Logistic回归模型在诊断PCa中具有较好的临床应用价值,可为临床准确诊断提供参考依据。Objective To construct a Logistic regression model based on multimodal ultrasound parameters for diagnosing prostate cancer(PCa),and to explore its clinical application value.Methods A total of 180 patients with suspected PCa admitted to our hospital were enrolled.According to pathological results,they were divided into PCa group(93 cases)and non-PCa group(87 cases).The prostate morphology,internal echo,envelope,nodules and calcification in two-dimensional ultrasound images and blood flow grading in CDFI were observed to obtain positive rates of two-dimensional ultrasound and CDFI images.The maximum,mean and minimum values of Young’s modulus(Emax,Emean,Emin)were obtained by shear wave elastography.The arrival time of contrast agent,time to peak,base and peak intensity,intensity difference and enhancement strength were obtained by contrast-enhanced ultrasound,and differences in the above examination results and clinical data between the two groups were compared.The independent influencing factors for predicting PCa were screened by multivariate Logistic regression analysis.The Logistic regression model was constructed,diagnostic efficiency of the model was analyzed by receiver operating characteristic(ROC)curve,and its goodness of fit was evaluated by Hosmer-Lemeshow test.A total of 32 patients with suspected PCa were enrolled as validation group for external verification.Results The levels of serum prostate specific antigen and prostate antibody density(PSAD)in PCa group were higher than those in non-PCa group(both P<0.05).The positive rates of two-dimensional ultrasound and CDFI images in PCa group were higher than those in non-PCa group(both P<0.05).Emax,Emean,Emin,peak intensity,intensity difference and enhancement strength in PCa group were higher than those in non-PCa group(all P<0.05),but there were no significant difference in other parameters between the two groups.Multivariate Logistic regression analysis showed that serum PSAD,Emax and intensity difference were independent influencing factors of PCa(a

关 键 词：超声检查 多普勒 彩色 造影剂 剪切波弹性成像 前列腺癌 

分 类 号：R445.1[医药卫生—影像医学与核医学]