多因素Logistic回归分析评价MSCT鉴别甲状腺良恶性结节的价值

Multi-factor Logistic regression analysis of MSCT in differential diagnosis of benign and malignant thyroid nodules

目的:探讨多因素Logistic回归分析评价MSCT鉴别甲状腺结节良恶性的价值。方法:选取87例经超声引导下细针穿刺活检或术后病理确诊的甲状腺结节,其中良性组44例,恶性组43例。以病理结果作为金标准评价MSCT诊断甲状腺结节良恶性的效能,并对良性组与恶性组结节的CT征象行Logistic回归分析。结果:单因素分析显示,形态不规则、边界中断征、细颗粒状钙化、与包膜分界不清、磨玻璃样强化、增强扫描边界不清是鉴别甲状腺良恶性结节的重要征象,2组差异均有统计学意义(均P<0.05);多因素回归分析显示,边界中断征、细颗粒状钙化、增强扫描边界不清与甲状腺癌相关(P=0.006,0.003,0.013)。结论:Logistic回归分析能有效提高MSCT诊断甲状腺恶性结节准确率,边界中断征、细颗粒状钙化及增强扫描边界不清与甲状腺癌密切相关,可作为辅助判断甲状腺结节良恶性的可靠指标。

Objective:To explore the value of MSCT in differentiating benign from malignant thyroid nodules by multi-factor Logistic regression analysis.Methods:A total of 87 patients with 104 thyroid nodules confirmed by ultrasound-guided fine needle aspiration or pathology were collected,among them 44 cases with 52 benign nodules became the benign group,and 43 cases with 52 malignant nodules became the malignant group.The efficiencies of MSCT in the diagnosis of benign and malignant thyroid nodules by taking pathological results as the gold standard were evaluated,and CT signs of thyroid nodules were analyzed by Logistic regression analysis.Results:One-way ANOVA analysis showed that irregular shape,boundary interruption sign,fine granular calcification,unclear boundary between nodule and thyroid capsule,ground glass enhancement and unclear boundary after enhancement were the signs of distinguishing benign and malignant thyroid nodules,the differences were statistically significant(all P<0.05).Multi-factor analysis showed that boundary interruption sign,fine granular calcification,unclear boundary after enhancement were correlated with thyroid cancer.Conclusions:Logistic regression analysis can effectively improve the accuracy of the diagnosis of thyroid malignant nodules.The boundary discontinuity,granular calcification and unclear boundary of contrastive scan are closely related to thyroid carcinoma and can be used as reliable indicators to help determine benign and malignant thyroid nodules.