分别采用甲状腺影像报告与数据系统（TI-RADS）与2015年美国甲状腺学会（ATA）《成人分化型甲状腺癌诊治指南》（简称ATA指南）推荐的超声特征对甲状腺结节进行恶性风险评估，并比较2种方法诊断甲状腺结节的价值。

选取2011年11月至2015年12月在北京协和医院获得病理结果的331例甲状腺结节患者，共485个结节，96个良性，389个恶性。评估结节大小、结构、纵横比、边缘、回声，钙化及腺体外侵犯。以TI-RADS及ATA指南分级标准对结节进行分级，比较各分级的恶性百分比。并以病理为"金标准"，取不同分级界值评估二者的诊断敏感度、特异度、阳性预测值、阴性预测值及准确性。采用受试者工作特征（ROC）曲线来确定TI-RADS与ATA分级的最佳临界值。并计算ATA指南与TI-RADS分级在最佳临界点诊断恶性时Kappa值。

TI-RADS分级2、3、4a、4b、4c及5级结节恶性百分比分别为0、12.0%（3/25）、22.2%（10/45）、29.8%（14/47）、99.2%（261/263）及100%（101/101）。ATA指南分级的良性、极低危、低危、中危、高危恶性百分比分别为0、12.5%（1/8）、16.1%（10/62）、27.7%（13/47）及99.2%（365/368）。2种方法各自分级之间恶性百分比比较，差异均有统计学意义（χ²=344.96、348.68，P均<0.01），TI-RADS分级与恶性百分比的相关系数为0.71，ATA指南分级相关系数为0.85。TI-RADS与ATA指南分级ROC曲线下面积分别为0.966、0.959，最佳分界点为TI-RADS分级≥4c与ATA指南分级≥高危，且当以≥4c与≥高危作为TI-RADS分级、ATA指南分级诊断恶性标准时，敏感度、特异度、阳性预测值、阴性预测值及准确性均数值较接近（93.1% vs 93.8%，97.9% vs 96.9%，99.5% vs 99.2%，77.7% vs 79.5%，94.0% vs 94.4%），差异均无统计学意义（P均>0.05），Kappa值=0.97。

TI-RADS与ATA分级对甲状腺结节的恶性风险评估诊断价值较高，以TI-RADS分级≥4c与ATA指南分级≥高危作为诊断恶性的标准时，二者诊断效率及一致性最高。

To compare diagnostic values of the 2015 American Thyroid Association (ATA) Management Guidelines for Adult Patients with Thyroid Nodules and Differentiated Thyroid Cancer with the thyroid imaging reporting and data system (TI-RADS) for sonographic malignancy risk stratification of thyroid nodules.

From November 2011 to December 2015, 485 thyroid nodules in 331 patients (mean age, 42.9 years±10.4)were included in this study. Characteristics includingsize, composition, shape(nonparallel or parallel), margin, echogenicity, calcifications and extrathyroidal extension of thyroid nodules were evaluated. Every nodule was stratificated by criteria set by TI-RADS and ATA guidelines, and malignant rate of each risk stratification were calculated and analysed. With pathology as the gold standard, different cutoff were taken to diagnose malignant nodules, and the sensitivity, specifity, positive predictive value, negativepredictive value and accuracy of the two methodologies were calculated at each cutoff. And the two methodologies were evaluated and measured by ROC curve.Finally their Kappa value were calculated at the best cutoff.

Of the 485 thyroid nodules, 96 were benign and 389 were malignant. The malignancy rates under TI-RADS category 2, 3, 4a, 4b, 4c, and 5 nodules were 0, 12.0% (3/25), 22.2% (10/45), 29.8% (14/47), 99.2% (261/363) and 100% (101/101). Malignancy rates under ATA guidelines of benign, very low, low, intermediate, and high suspicion for malignancy were 0,12.5% (1/8), 16.1% (10/62), 27.7% (13/47), and 99.2% (365/368). There were significant differences inside each patterns (P<0.01) respectively and high correlation between risk stratification with TI-RADS (r=0.70) and ATA guidelines (r=0.83). Areas under the ROC curve of the TI-RADS and ATA guidelines classifications were 0.966 and 0.959. Best cut-off point for diagnosing malignant by TI-RADS and ATA guideline classifications were ≥4c and ≥high suspicion, and at that point, diagnostic value of TI-RADS and ATA guidelines were nearly the same(sensitivity, 93.1% vs 93.8%; specificity, 97.9% vs 96.9%; PPV, 99.5% vs 99.2%; NPV, 75.7%vs 79.5%; and accuracy, 94.0%vs 94.4%), and there was no significant differences (P=0.50, P=0.50, P=0.50, P=0.53, P=0.55), Kappa=0.97.

Both TI-RADS and the ATA guidelinesprovide effective malignancy risk stratification for thyroid nodules. The diagnosticvalue of TI-RADS when considering≥4c and ATA guidelines when considering ≥high-suspicion nodules as malignant were nearly the same and both high.