基于经颅超声图像特征的Logistic回归模型预测帕金森病的临床价值  

Clinical value of Logistic regression model based on transcranial ultrasound imaging features in predicting Parkinson’s disease

作　　者：周紫薇 方靖琴 胡加银 田优幽 蒙云 李陶 ZHOU Ziwei;FANG Jingqin;HU Jiayin;TIAN Youyou;MENG Yun;LI Tao(Department of Ultrasound,Daping Hospital,Army Medical University,Chongqing 400020,China)

机构地区：[1]陆军军医大学大坪医院超声科,重庆400020

出　　处：《临床超声医学杂志》2025年第2期142-146,共5页Journal of Clinical Ultrasound in Medicine

基　　金：重庆市技术创新与应用示范社会民生类一般项目(cstc2018jscx-msybX0018)。

摘　　要：目的应用经颅超声观察帕金森病患者中脑黑质、豆状核、中缝核回声变化,并基于此构建Logistic回归模型,探讨其在诊断帕金森病中的临床价值。方法选取我院收治的231例疑似帕金森病患者,其中原发性帕金森病患者108例(PD组),非原发性帕金森病患者123例(非PD组),应用经颅超声观察中脑黑质、豆状核、中缝核回声特点,比较两组超声图像特征及临床资料的差异;应用二元Logistic回归分析筛选预测帕金森病的独立影响因素,并基于此构建模型。绘制受试者工作特征(ROC)曲线、临床决策曲线分析该模型对帕金森病的诊断效能及临床净获益。结果两组年龄、黑质强回声、豆状核回声强度、中缝核连续性比较,差异均有统计学意义(均P<0.05);性别、临床症状(震颤/肌强直、运动迟缓)比较差异均无统计学意义。二元Logistic回归分析显示,年龄、黑质强回声、豆状核回声强度、中缝核连续性均为预测帕金森病的独立影响因素(均P<0.05)。基于上述4个变量构建回归模型为:Logit(P)=-7.338+0.038×年龄+0.991×黑质强回声+1.076×豆状核回声强度+1.765×中缝核连续性。ROC曲线分析显示,当该模型截断值为0.393时预测帕金森病的曲线下面积为0.851,灵敏度为79.6%,特异度为80.5%,约登指数为0.601。临床决策曲线分析显示,当概率阈值为0.1~1.0时该模型有较好的临床净获益。结论基于经颅超声图像特征的Logistic回归模型可用于辅助诊断帕金森病,且具有较好的临床应用价值。Objective To observe the echogenic changes of the substantia nigra,lenticular nucleus and raphe nucleus in the midbrain in patients with Parkinson’s disease by transcranial ultrasound,and a Logistic regression model was constructed,to explore the clinical value of the model in the diagnosis of Parkinson’s disease.Methods A total of 231 patients with suspected Parkinson’s disease admitted to our hospital were selected,including 108 patients with primary Parkinson’s disease(PD group)and 123 patients with non-primary Parkinson’s disease(non-PD group),and transcranial ultrasound was used to observe the echo characteristics of substantia nigra,lenticular nucleus and raphe nucleus of the midbrain,the differences in the above ultrasound imaging features and clinical between the two groups were compared.Binary Logistic regression analysis was used to screen the predictive independent influencing factors for predicting Parkinson’s disease,and a model was constructed.The diagnostic efficacy of the model for Parkinson’s disease and the net clinical benefit were analyzed by receiver operating characteristics(ROC)curve and clinical decision curve.Results There were significant differences in age,nigrostriatal strong echo,intensity of echoes in the lenticular nucleus and continuity of the raphe nucleus between the two groups(all P<0.05).There were no significant difference in gender and clinical symptoms(tremor/myotonia and bradykinesia).Binary Logistic regression analysis showed that age,nigrostriatal strong echo area,intensity of echoes in the lenticular nucleus and continuity of the raphe nucleus were independent influencing factors for predicting Parkinson’s disease(all P<0.05).A regression model was constructed based on the above 4 variables:Logit(P)=-7.338+0.038×age+0.991×nigrostriatal strong echo+1.076×intensity of echoes in the lenticular nucleus+1.765×continuity of the raphe nucleus.ROC curve analysis showed that when the cutoff value of the model was 0.393,the area under the curve for predicting o

关 键 词：超声检查 经颅 帕金森病 黑质 中缝核 豆状核 

分 类 号：R445.1[医药卫生—影像医学与核医学]