探讨2017年美国放射学会（ACR）提出的甲状腺超声影像报告和数据系统（TI-RADS）鉴别诊断甲状腺结节良恶性的效能及其应用价值。

回顾性分析2016年11月及12月东部战区总医院收治的经细针穿刺活检（FNA）或手术病理证实的甲状腺结节患者176例，共216个结节。根据TI-RADS分类对结节的超声特征（包括结构、回声、形态、边缘、局灶性强回声）量化评分，根据结节大小将结节分为FNA组、随访组、不予处理组。采用χ²检验比较TI-RADS分类鉴别诊断甲状腺结节良恶性与病理结果差异。以病理结果作为"金标准"，绘制TI-RADS分类鉴别诊断甲状腺结节良恶性的受试者操作特征（ROC）曲线。

216个甲状腺结节中，良性84个，恶性132个。TI-RADS 1、2、3、4、5类结节分别为3、24、23、60、106个，恶性率分别为0、0、17.3%（4/23）、48.3%（29/60）、93.4%（99/106）。以TI-RADS 1~3类判为良性，TI-RADS 4、5类判为恶性，TI-RADS分类鉴别诊断甲状腺结节良恶性与病理结果比较，差异有统计学意义（χ²=77.222，P<0.001）。TI-RADS分类鉴别诊断甲状腺结节良恶性的敏感度、特异度、阳性预测值、阴性预测值分别为97.0%、54.2%、77.2%、91.8%。ROC曲线显示，TI-RADS分类鉴别诊断甲状腺结节良恶性的最佳阈值为5.5分，曲线下面积（AUC）为0.918，标准误为0.019，95%CI为0.881~0.956，约登指数为0.699。TI-RADS 1、2类结节均为不予处理组，良性率为100.0%（3/3，24/24）；3类结节FNA组、随访组、不予处理组恶性率分别为18.2%（2/11）、0、25.0%（2/8）；4类结节3组恶性率分别为31.0%（9/29）、63.6%（7/11）、65.0%（13/20）；5类结节3组恶性率分别为96.3%（52/54）、88.9%（40/45）、100.0%（7/7）。

TI-RADS分类能有效鉴别甲状腺结节良恶性，尤其是对甲状腺良性结节，有较高的诊断能力。TI-RADS 3类及以上不同大小的甲状腺结节均有恶性可能，应用TI-RADS分类判断可疑恶性的甲状腺结节是否行FNA时可放宽大小指征。

To assess the value of the thyroid imaging reporting and data system (TI-RADS), which was developed by the American College of Radiology in 2017, in diagnosing benign and malignant thyroid nodules.

A total of 176 patients with 216 thyroid nodules were retrospectively collected in this study at General Hospital of Eastern Theater Command. They were quantitatively evaluated for ultrasound features (structure, echo, morphology, margins, and focal echogenicity) by using the 2017 TI-RADS ultrasonographic stratification and were divided into a fine needle aspiration (FNA), a follow-up group, and an undealt group according to nodule size. The results were compared with FNA or surgery-based pathological results. The results of TI-RADS stratification between benign and malignant thyroid nodules were compared by the chi-square test. The receiver operating characteristic (ROC) curve analysis was conducted to determine the diagnostic values of thyroid TI-RADS stratification.

There were 84 benign and 132 malignant nodules in 216 thyroid nodules. They were classified into five categories according to the scores 0-14. The malignant rates of TI-RADS 1, 2, 3, 4, and 5 were 0, 0, 17.3%(4/23), 48.3%(29/60), and 93.4%(99/106), respectively. When TI-RADS 1, 2, and 3 thyroid nodules were classified as benign nodules, and TI-RADS 4 and 5 categories were classified as malignant nodules, the sensitivity, specificity, positive predictive value, and negative predictive value were 97.0%, 54.2%, 77.2%, and 91.8%, respectively. The area under the ROC curve was 0.918 (95% confidence interval [CI]: 0.881-0.956), and the Youden index was 0.699 with a critical value of 5.5. There was a statistically significant difference between TI-RADS stratification and pathological examination (χ²=77.222, Ρ<0.001). TI-RADS 1 and 2 nodules belonged to the undealt group. The rates of malignancy in TI-RADS 3 nodules were 18.2%(2/11), 0, and 25.0%(2/8) in the FNA group, follow-up group, and undealt group, respectively; the corresponding rates in TI-RADS 4 and 5 nodules were 31.0%(9/29), 63.6%(7/11), and 65.0%(13/20), and 96.3%(52/54), 88.9%(40/45), and 100.0%(7/7), respectively.

The TI-RADS classification of thyroid nodules has a high accuracy to identify benign and malignant nodules, especially benign nodules. Since TI-RADS 3 and above nodules of different sizes are all likely to be malignant, the indications for FNA of suspected malignant thyroid nodules according to TI-RADS classification in terms of nodule size may not be strictly obeyed.