基于MRI平扫征象预测腮腺肿瘤良恶性 列线图模型的构建与应用  

Construction and application of a nomogram for predicting benign and malignant parotid gland tumors based on convention MRI

作　　者：郭昊 史灵雪 刘硕[3] 于保婷 管玉瑶 郑雪微 李佟[3] 杨鸿声 冯婧[4] GUO Hao;SHI Lingxue;LIU Shuo;YU Baoting;GUAN Yuyao;ZHENG Xuewei;LI Tong;YANG Hongsheng;FENG Jing(Department of Radiology,China-Japan Union Hospital of Jilin University,Changchun 130003,China)

机构地区：[1]吉林省长春市人民医院放射线科,吉林长春130033 [2]吉林省人民医院放射线科,吉林长春130021 [3]吉林大学中日联谊医院放射科,吉林长春130003 [4]吉林大学口腔医院放射科,吉林长春130021

出　　处：《中国中西医结合影像学杂志》2023年第4期397-401,406,共6页Chinese Imaging Journal of Integrated Traditional and Western Medicine

基　　金：吉林省教育厅科研基金资助课题(JJKH20201057KJ)。

摘　　要：目的:基于MRI平扫征象建立腮腺肿瘤良恶性鉴别的预测模型,为腮腺肿瘤的术前诊治提供重要依据。方法:选取经病理确诊的138例腮腺肿瘤患者为建模组,通过logistic回归分析筛选出腮腺肿瘤良恶性鉴别的独立影响因素,建立肿瘤良恶性数学预测模型;另收集经病理确诊的腮腺肿瘤患者35例进行验证(验证组)。结果:单因素和多因素logistic回归分析显示,肿瘤形态、边界、信号均匀性、ADC值这4个指标是预测腮腺肿瘤良恶性的独立影响因素(均P<0.05)。通过对建模组行多因素logistic回归分析,建立数学预测模型如下:Y=eX/(1+eX),X=0.385+(-1.416×肿瘤形态)+(-1.473×肿瘤边界)+(-1.306×肿瘤信号)+(2.312×肿瘤ADC值)。结果显示,该模型对腮腺肿瘤良恶性判断ROC曲线的AUC为0.832(95%CI 0.75~0.91),敏感度为82.6%,特异度为70.7%,准确率为70.7%;利用验证组验证该模型,AUC为0.936(95%CI 0.83~0.98),敏感度为85.7%,特异度为96.4%,准确率为94.3%。结论:联合肿瘤形态、边界、信号、ADC值建立的预测模型可为临床术前鉴别腮腺肿瘤良恶性提供重要参考依据。Objective:A prediction model of benign and malignant differentiation was established by MRI signs and basic clinical data on parotid gland tumors to provide an important basis for the preoperative diagnosis and treatment of parotid gland tumor patients.Methods:The data of 138 patients(the modeling group)diagnosed by pathology were retrospectively analyzed.The independent influencing factors for benign and malignant differentiation of parotid tumors were selected by logistic regression analysis,and a mathematical prediction model for benign and malignant tumors was established.The data of 35 patients diagnosed by pathology were collected as the validation group.Results:Univariate and multivariate logistic regression analysis showed that tumor morphology,border,signal and ADC value were independent risk factors for predicting benign and malignant parotid gland tumors(all P<0.05).Through multivariate logistic regression analysis on the modeling group,the mathematical prediction model was established as follows:Y=eX/(1+eX),X=0.385+(-1.416×morphology)+(-1.473×border)+(-1.306×signal)+(2.312×ADC value).The results showed that the AUC of the model was 0.832(95%CI 0.75~0.91),the sensitivity was 82.6%,the specificity was 70.7%,the accuracy rate was 70.7%.And in the validation group,the AUC was 0.936(95%CI 0.83~0.98),the sensitivity was 85.7%,the specificity was 96.4%,and the accuracy rate was 94.3%.Conclusion:Combined with tumor morphology,border,signal and ADC value,the established predictive model can provide an important reference for the preoperative diagnosis of benign and malignant parotid gland tumors.

关 键 词：腮腺肿瘤 LOGISTIC回归 预测模型 列线图 

分 类 号：R739.87[医药卫生—肿瘤] R445.2[医药卫生—临床医学]