摘要
目的应用Logistic回归模型评价甲状腺囊实性结节的超声检查各特征指标对良恶性结节鉴别诊断的意义。方法回顾分析手术病理证实的136个甲状腺囊实性结节(良性结节89个,恶性结节47个)的术前的灰阶超声和彩色多普勒的各项超声诊断指标,建立Logistic回归模型,并评价其预报能力。结果共有4个特征进入Logistic回归模型,分别是囊实性结节内钙化、实性部分位置、边缘、内部结构。Logistic回归模型对甲状腺囊实性结节良恶性预报的准确率为94.1%(128/136),ROC曲线下面积为0.930±0.029。结论超声特征的Logistic回归模型有助于甲状腺囊实性结节的良恶性鉴别诊断。
Objective To evaluate the application of Logistic regression model in analyzing sonographic fea- tures of Partially cystic thyroid nodules. Methods The indexes of two-dimensional uhrasonography and color Doppler flow imaging of 89 benign and 47 malignant Partially cystic thyroid nodules were retrospectively reviewed, which were confirmed by pathological diagnosis after surgery.A Logistic model for predicting malignant of Partially cystic thyroid nodule was obtained, and assessment of the predicting value was performed. Results Four sonographic characteristics were finally entered into the Logistic regression model including calcification type, margin, position of solid portion, internal content of Partially cystic thyroid nodules. The percentage correction of prediction was 94.1% （128/136）. The area under curve was 0.930±0.029. Conclusion The Logistic model of sonographic features can help differentiate benign from malignant Partially cystic thyroid nodules.
出处
《实用医学影像杂志》
2017年第4期289-291,共3页
Journal of Practical Medical Imaging
