Objective: To summarize and evaluate the value of applying the thyroid imaging reporting and data system (TI-RADS) released by American College of Radiology (ACR) in 2017 of the thyroid classification, and to propose an optimized classification method based on the result to facilitate more accurate and precise risk stratification of thyroid nodules.
Methods: In the study, 342 thyroid nodules assessed by 2017 ACR TI-RADS were retrospectively analyzed. Each nodule had a score, and all the scores of nodules were compared with the pathological results. The proportion of malignant nodules in different scoring ranges was obtained. The diagnostic efficacy of all nodules, nodules above 1 cm and less than or equal to 1 cm was evaluated by ROC curve, respectively.
Results: The AUC of all nodules, nodules above 1 cm and less than or equal to 1 cm were 0.907, 0.936 and 0.717, respectively. With the increase of the scores, the proportion of benign nodules decreased gradually, and the proportion of malignant nodules increased, especially nodules of 4-6 scores increased significantly. Based on the proportion of malignant nodules with 3 scores, the proportion of malignant nodules with 4, 5 and 6 scores increased 1.6, 3.8 and 5.3 times, respectively. The proportion of malignant nodules with 6-8 scores was 81%-84%, while the proportion of malignant nodules with 9 scores or more was 93%-94%. According to the distribution characteristics of malignant nodules, the classification of TI-RADS was adjusted. TI-RADS 4 was divided into TI-RADS 4a, TI-RADS 4b and TI-RADS 4c, corresponding to 4, 5 and 6-8 scores respectively, while the nodules with 9 scores or more were divided into TI-RADS 5.
Conclusion: 2017 ACR TI-RADS has high diagnostic value for thyroid nodules above 1 cm, but it is not so effective for the nodules less than or equal to 1 cm. According to the proportion distribution of malignant nodules in different scoring ranges, appropriate adjustment of classification will be more accurate and precisely predict the malignant risk of nodules.
目的: 评估应用2017年美国放射学会(American College of Radiology,ACR)发布的甲状腺影像学报告与数据系统(thyroid imaging reporting and data system,TI-RADS) 对甲状腺进行危险分层的价值,并依据结果提出优化分类的建议。
方法: 回顾性分析北京大学第三医院应用2017版ACR TI-RADS评估的342例影像资料完整的甲状腺结节,将评分结果与病理结果进行对比,获得不同分值区间恶性结节的比例,并分别对最大径>1 cm及最大径≤1 cm的结节使用ROC曲线评价诊断效能。
结果: 利用该评分系统对结节进行危险分层,全部结节、最大径>1 cm的结节、最大径≤1 cm的结节ROC曲线下面积分别为0.907、0.936、0.717。随着评分值的增加,良性结节比例逐渐下降,恶性结节所占比例逐渐增长,评分值4~6分区间恶性结节比例增长明显,以评分值为3的恶性结节比例为基准,4、5、6分结节恶性结节分别增长1.6倍、3.8倍、5.3倍,6~8分区间恶性结节稳定在81%~84%,而9分及以上恶性结节比例稳定在93%~94%,依据恶性结节的比例分布特点调整分类,TI-RADS 1类、TI-RADS 2类、TI-RADS 3类仍然分别对应0分、2分、3分,TI-RADS 4类细分为TI-RADS 4a类、TI-RADS 4b类、TI-RADS 4c类,分别对应4分、5分、6~8分,而≥9分的结节划分为TI-RADS 5类。
结论: 2017版ACR TI-RADS对最大径>1 cm的甲状腺结节具有较高的诊断价值,而对最大径≤1 cm的结节诊断价值欠佳。根据不同评分值区间恶性结节比例的分布特点,适当调整分类将能更详细、准确地预测结节的恶性风险。