基于超声指标的甲状腺结节良恶性Nomogram预测模型的建立与验证  

Establishment and Validation of A Nomogram Predictive Model for Benign and Malignant Thyroid Nodules Based on Ultrasound Indicators

作　　者：阿衣尔红 陈志远 韩立明 周文秀 魏星 张旭胜 AYI Erhong;CHEN Zhiyuan;HAN Liming;ZHOU Wenxiu;WEI Xing;ZHANG Xusheng(Department of Ultrasound,the Third Affiliated Hospital of Chengdu Medical College(Chengdu Pidu District People's Hospital),Chengdu Sichuan 611730,China)

机构地区：[1]成都医学院第三附属医院(成都市郫都区人民医院)超声科,四川成都611730 [2]成都医学院第三附属医院(成都市郫都区人民医院)检验科,四川成都611730

出　　处：《联勤军事医学》2024年第8期672-677,共6页Military Medicine of Joint Logistics

基　　金：成都市医学科研课题项目(2021139)。

摘　　要：目的 建立基于超声指标的甲状腺结节良恶性Nomogram预测模型并进行验证。方法 收集2021-04/2023-07月在作者医院行甲状腺结节手术、病历完整且术后甲状腺结节良恶性病理诊断明确的423例患者的临床资料。根据入院时间将患者分为模型组(n=296)和验证组(n=127)。收集所有患者的一般资料、常规超声、实时弹性成像与超声造影检查结果。根据病理检查结果将模型组患者分为良性组(n=93)与恶性组(n=203),比较两组患者的一般资料、常规超声、实时弹性成像及超声造影指标,以最小绝对值收敛和选择算子算法(least absolute shrinkage and selection operator,LASSO)回归筛选潜在变量后行多因素Logisitic回归,然后建立Nomogram预测模型并进行验证。结果 多因素Logistic回归分析结果显示:年龄、结节数目、结节纵横比、结节边界、平均弹性值(elasticity mean,Emean)、均匀度、峰值强度(peak intensity,PI)、曲线下面积(area under the curve,AUC)为甲状腺结节良恶性的独立影响因素(P均<0.05)。根据多因素回归分析结果,以R软件建立列线图模型。受试者工作特征(receiver operating characteristic,ROC)曲线分析结果显示模型组列线图预测甲状腺结节良恶性的AUC为0.836,95%置信区间(confidence interval,CI)为0.787~0.886,特异度为68.80%,灵敏度为81.80%;验证组列线图预测甲状腺结节的良恶性的AUC为0.808,95%CI为0.752~0.863,特异度为72.00%,灵敏度为82.10%。校准曲线分析结果显示:模型组与验证组的预测曲线与标准曲线基本拟合。决策曲线结果显示当模型预测值为18.20%~88.50%时患者净获益率>0。结论 甲状腺结节的良恶性与年龄、结节数目、结节纵横比等因素有关,基于超声指标的Nomogram预测模型可预测甲状腺结节的良恶性。Objective To establish a Nomogram predictive model for benign and malignant thyroid nodules based on ultrasound indicators and validate it.Methods The clinical data of 423 patients who underwent thyroid nodule surgery in author′s hospital from April 2021 to July 2023,with complete medical records and clear postoperative pathological diagnosis were selected.The patients were divided into model group(n=296) and validation group(n=127) according to admission time.General information,routine ultrasound,real-time elastography and contrast-enhanced ultrasound indicators of those patients were collected.The model group patients were divided into benign group(n=93) and malignant group(n=203) according to the pathological examination results,the general information,routine ultrasound,real-time elastography and contrast-enhanced ultrasound indicators of the two groups of patients were compared,potential variables were screened using least absolute shrinkage and selection operator(LASSO) regression and multivariate Logistic regression were performed,and a Nomogram predictive model was established and validated.Results Among the 296 patients in the model group,203 were diagnosed with malignant tumors through pathological examination.The results of multivariate Logistic regression analysis showed that age,number of nodules,aspect ratio,boundary,elasticity mean(Emean),uniformity,peak intensity(PI) and area under the curve(AUC) were independent influencing factors for benign and malignant thyroid nodules(all P<0.05).Column chart model was established using R software based on the results of multi factor regression analysis.The receiver operating characteristic(ROC) curve analysis results showed that the AUC of the model group′s nomogram for predicting benign and malignant thyroid nodules was 0.836,95% confidence interval(CI) was 0.787-0.886 specificity was 68.80% and sensitivity was 81.80%;the AUC of the validation group′s column chart for predicting benign and malignant thyroid nodules was 0.808,95% CI was 0.752-0.863,sp

关 键 词：超声指标 甲状腺结节 良性 恶性 甲状腺癌 多因素分析 Nomogram预测模型 

分 类 号：R736.1[医药卫生—肿瘤]