甲状腺实性结节超声特征鉴别诊断的Logistic回归分析  

Logistic regression analysis of ultrasonic characteristics in differential diagnosis of thyroid solitary nodules

作　　者：侯秀敏[1] 王荣[1] 王兴田[1] 

机构地区：[1]徐州医学院附属医院超声医学科,江苏徐州221002

出　　处：《徐州医学院学报》2015年第1期35-38,共4页Acta Academiae Medicinae Xuzhou

基　　金：徐州市科技项目（XZZD1337）

摘　　要：目的应用Logistic回归模型对甲状腺实性结节的超声特征进行分析，评价常规超声及声触诊组织成像（VTI）在甲状腺实性结节良恶性鉴别诊断中的价值。方法对129例甲状腺实性结节患者（152枚实性结节）的二维、彩色血流及VTI检查图片进行回顾性分析，以病理结果为金标准，建立Logistic回归模型，比较进入方程中的变量的优势比（OR），评价各变量的鉴别诊断效能。结果经过后退法Logistic回归分析，共筛选出3个对甲状腺实性结节良恶性鉴别有统计学意义的特征变量，分别为边界、微钙化、VTI评分。Logistic回归模型对甲状腺实性结节良恶性预报的ROC曲线下面积为0．972±0．014。结论二分类Logistic回归模型分析能筛选出对甲状腺实性结节鉴别有诊断意义的特征变量。联合应用常规超声及VTI声像图特征对甲状腺实性结节的鉴别诊断具有较高的临床煮义．Objective To analyze the ultrasonic characteristics of thyroid solid nodules using logistic regression and to evaluate the application of conventional uhrasonograpby and virtual touch tissue imaging （VTI） in the differential diag- nosis of benign and malignant solid thyroid nodules. Methods Clinical data from 129 patients with thyroid solid nodules （totaling 152 nodules） were retrospectively analyzed, including 2D ultrasound, color Doppler flow imaging （CDFI） and VTI. Based on pathological results as golden diagnosis criteria, a logistic model was established. Then, the effectiveness of differential diagnosis of each variable was assessed through comparison of its odds ratio （OR）. Results Through logistic retrospective analysis, three variables with statistically significance were finally screened out, detailing edge, microcalcification and VT[ score. Based on the logistic model, the area under ROC curve which could predict benign and malignant solid thyroid nodules was 0.972 ± 0. 014. Conclusion The binary logistic regression model can screen out the variables useful for differential diagnosis of benign and malignant solid thyroid nodules. Combined use of conventional ultrasonography and VTI play a critial role in the differential diagnosis of solid thyroid nodules.

关 键 词：超声 甲状腺 实性结节 LOGISTIC模型 声触诊组织成像 

分 类 号：R445.1[医药卫生—影像医学与核医学]