ACR TI-RADS分类中各超声特征对甲状腺恶性结节的诊断效能

The diagnostic efficacy of various ultrasound features in ACR TI-RADS classification for malignant thyroid nodules

目的探讨2017年美国放射学会甲状腺影像报告和数据系统(ACR TI-RADS)中各超声特征在甲状腺恶性结节中的诊断效能。方法回顾性分析2018年6月~2020年5月我院诊断为甲状腺结节患者的临床资料。1004个结节均于超声检查后接受FNA细胞学检查或外科手术治疗确诊。用ACR TI-RADS方法对上述结节进行分类评分,并与术后病理对照,评价其各超声特征诊断效能。结果ACR TI-RADS分类准确度为77.0%,灵敏度为77.8%,特异度为76.1%。赋值评分的AUC(0.828)高于边缘(0.792)和回声(0.780),回声与边缘相比,回声灵敏度高于边缘(91.3%vs 68.0%),但特异度低于边缘(64.1%vs 89.1%),差异均有统计学意义(P<0.05)。低回声、极低回声患者病理结果为恶性的风险分别为53.043倍和72.455倍;边缘分叶、不规则者的风险分别为3.789倍和9.132倍。结论ACR TI-RADS分类中赋值评分在甲状腺恶性结节中诊断效能最高,认为TR5级以上恶性的风险会更高;极低回声、低回声,边缘不规则、分叶特征相对其他特征具有较高的诊断效能。

Objective To explore the diagnostic efficacy of each ultrasound feature in the Thyroid imaging reporting and data system of ACR(ACR TI-RADS)in 2017 in the diagnosis of malignant thyroid nodules.Methods The clinical data of patients diagnosed with thyroid nodule in our hospital from June 2018 to May 2020 were retrospectively analyzed.All 1004 nodules were confirmed by FNA cytology or surgical treatment after ultrasound examination.ACR TI-RADS was used to classify and score the nodules,and the diagnostic efficacy of each ultrasound feature was evaluated by comparing with postoperative pathology.Results The classification accuracy,sensitivity and specificity of ACR TI-RADS were 77.0%,77.8%and 76.1%respectively.The AUC(0.828)of the assigned score was higher than that of the edge(0.792)and the echo(0.780).Compared with the edge,the echo sensitivity was higher than that of the edge(91.3%vs 68.0%),but the specificity was lower than that of the edge(64.1%vs 89.1%).The risk of malignancy in hypoechoic and extremely hypoechoic patients was 53.043 and 72.455 times,respectively.The risk of those with unclear boundary and irregular edge were 3.789 and 9.132 times,respectively.Conclusion The evaluation score in the ACR TI-RADS classification is the highest in the diagnosis of malignant thyroid nodules,and it is considered that the risk of malignancy above TR5 will be higher;very low echo,low echo,irregular edges,and lobulated features have higher relative to other features Diagnostic efficacy.