甲状腺结节CT特征的Logistic回归分析  

Logistic regression analysis of CT features in thyroid nodule

作　　者：解丙坤[1] 李子浩[1] 周建立[1] 王敏[2] 洒盼盼[1] 李洪雷[1] 贾守强[3] 

机构地区：[1]德州市人民医院放射科,山东德州253014 [2]淄博市中心医院放射科 [3]莱芜市人民医院影像科

出　　处：《实用放射学杂志》2015年第11期1765-1769,共5页Journal of Practical Radiology

摘　　要：目的 应用Logistic回归分析探讨并筛选甲状腺结节良恶性的CT特征指标.方法 回顾性分析经病理证实的98例甲状腺结节患者,记录良恶性结节的CT特征并进行多因素回归分析,建立Logistic回归模型.结果 98例结节中良性结节112个,恶性48个.经单因素分析,病变的数目、形态、边界、钙化、囊变、强化方式、腺外侵犯及淋巴结肿大差异有统计学意义,将有意义的CT特征纳入Logistic回归分析,病灶边界、钙化、淋巴结肿大3个特征变量有意义.结论 病变边界、钙化及淋巴结肿大是鉴别甲状腺结节性病变良恶性的3个特征性预测指标,Logistic回归模型有助于鉴别甲状腺良恶性结节.Objective To evaluate and screen out the CT characteristic features of benign and malignant thyroid nodules by using the Logistic regression analysis. Methods 98 cases with thyroid nodules diagnosed by pathology were analyzed retrospectively. The CT features of benign and malignant nodules were recorded and compared by using multiple stepwise Logistic regression analysis. The Logistic regression model was established. Results There were 98 thyroid nodules,including 112 benign and 48 malignant nodules. There were significant differences in the number, shape, border, calcification, cystic degeneration, enhancement manner, periglandular invasion and lymphadenectasis between benign and malignant nodules by using single factor analysis. The border, calcification and lymphadenectasis were significant features by using Logistic regression analysis. Conclusion The border, calcification and lymphadenectasis are the independent characteristic CT features to predict benign or malignant thyroid nodules. The Logistic regression model can be helpful for differentiation of benign or malignant thyroid nodules.

关 键 词：甲状腺结节 计算机体层成像 LOGISTIC回归分析 

分 类 号：R581[医药卫生—内分泌] R814.42[医药卫生—内科学]