Logistic regression analysis of multi-mode ultrasonography in the diagnosis of thyroid carcinoma
10.3877/cma.j.jssn.2095-5820.2016.12.009
- VernacularTitle:多模式超声检查诊断甲状腺癌患者的Logistic回归分析
- Author:
Zhenpeng, LENG
;
Jingchun, YANG
;
Huijuan, MA
;
Jia, MA
;
Meijie, REN
;
Ying, CHANG
;
Hua, KANG
;
Lianghong, TENG
- Keywords:
Thyroid neoplasms;
Ultrasongraphy;
Elasticity imaging techniques
- From:
Chinese Journal of Medical Ultrasound (Electronic Edition)
2016;13(12):918-924
- CountryChina
- Language:Chinese
-
Abstract:
Objective To evaluate the diagnostic value of gray-scale ultrasonography (GSU) combined with contrast-enhanced ultrasonography (CEUS) and real-time elastography imaging (EI) in solid thyroid nodules.Methods A total of 124 solid nodules were collected from Xuanwu Hospital of Capital Medical University from September 2011 to November 2014, including 49 benign nodules and 75 malignant nodules conifrmed by surgical pathology. The data of multimodal imaging were retrospectively analyzed and compared with the pathological results. The parameters among the groups were compared byχ2 test. Four diagnostic models including the statistically signiifcant independent variables were established accord to different combinations of routine ultrasound examination. The binary Logistic regression equation using the pathological results as the dependent variable Y and the ultrasound features of thyroid nodules as independent variableX was established for calculating the area under the receiver operator curve (ROC) and multiple regression analysis furthermore.Results There were statistically significant differences in aspect ratio, margin condition, echo type, CEUS features, elastic imaging score and elastic strain rate of benign and malignant thyroid nodules(χ2=13.966, 7.106, 24.215, 17.250 and 27.158,P<0.05).The four Logistic regression models were statistically signiifcant (χ2=5.37, 5.37, 8.155, 8.155, allP<0.05). By using modelⅠ,Ⅱ,Ⅲ andⅣ, the diagnostic sensitivity, speciifcity and accuracy of malignant nodules were (73.8%, 75.9%, 79.7%, 83.3%), (67.5%, 70.7%, 85.2%, 86.7%), and (71.8%, 74.2% 81.8%, 84.4%) respectively. The area under the ROC of logistic equation predicted values were (0.778, 0.809, 0.875 and 0.898, allP<0.01). Multivariate categorical analysis showed that aspect ratio, margin condition and the elastic strain rate ratio (χ2=13.73, 7.541 and 11.461, allP<0.05) of thyroid nodules were effective diagnostic indicators. Conclusions Our study indicated that CEUS and EI could be helpful modality for GSU to improve the diagnostic accuracy of thyroid nodules. The combined application of multi mode ultrasound examination has a high diagnostic value.
