摘要目的:探讨C-TIRADS联合人工智能辅助诊断S-Detect技术在甲状腺结节鉴别诊断中的诊断价值。方法:回顾性分析2020年4-9月在河南省肿瘤医院进行超声检查并明确病理结果的237例甲状腺结节患者(237个结节)，按照C-TIRADS指南标准对结节进行分类诊断，然后使用S-Detect技术联合C-TIRADS对结节再次分类诊断，以病理结果为金标准绘制ROC曲线，比较二者诊断结果的ROC曲线下面积、敏感性、特异性和准确性。结果:237个甲状腺结节中良性结节105个，恶性结节132个。单独使用C-TIRADS诊断及联合人工智能C-TIRADS诊断的ROC曲线下面积分别为0.869及0.942，两组间差异有统计学意义(χ 2＝36.11， P<0.001)；以C-TIRADS 4A类作为良恶性结节鉴别诊断的标准时，联合人工智能辅助诊断的C-TIRADS分类较单独使用C-TIRADS诊断的特异性及准确性明显提高，差异有统计学意义(83.81%对47.62%，90.72%对75.53%，均 P<0.05)。 结论:C-TIRADS联合人工智能辅助诊断S-Detect技术具有较高的诊断效能，可以提高甲状腺结节诊断的特异性和准确性，减少不必要的穿刺活检。

abstractsObjective:To explore the diagnostic value of C-TIRADS combined with artificial intelligence-assisted diagnosis S-Detect technology in the differential diagnosis of thyroid nodules.Methods:A total of 237 thyroid nodules patients (237 thyroid nodules)with ultrasound examination and definitive pathologic results in Henan Cancer Hospital from April to September 2020 were retrospectively analyzed. The nodules were diagnosed according to C-TIRADS guidelines, and then by S-Detect technology combined with C-TIRADS guidelines. The ROC curve was plotted with the pathological results as the gold standard, and the area under the ROC curve, sensitivity, specificity and accuracy of the diagnosis results between the two groups were compared.Results:Among the 237 thyroid nodules, 105 were benign and 132 were malignant.The area under the ROC curve of C-TIRADS diagnosis alone and C-TIRADS diagnosis combined with artificial intelligence were 0.869 and 0.942 respectively, the difference between the two groups was statistically significant (χ 2=36.11, P<0.001); When Category 4A was used as the cutoff value of benign and malignant differential diagnosis, the specificity and accuracy of C-TIRADS classification of artificial intelligence-assisted diagnosis was significantly higher than that of C-TIRADS alone, and the difference was statistically significant(83.81% vs 47.62%, 90.72% vs 75.53%, all P<0.05). Conclusions:C-TIRADS combined with artificial intelligence-assisted diagnosis S-Detect technology has a high efficiency in the diagnosis of thyroid nodules and can improve the specificity and accuracy of thyroid nodules diagnosis and reduce unnecessary biopsy.