分子影像学杂志2026，Vol.49Issue(5)：604-610,7.DOI:10.12122/j.issn.1674-4500.2026.05.06
ACR-TIRADS、C-TIRADS与AI-TIRADS在甲状腺结节良恶性诊断中的应用效果
The application effects of ACR-TIRADS,C-TIRADS and AI-TIRADS in the diagnosis of benign and malignant thyroid nodules
摘要
Abstract
Objective To compare the diagnostic efficiency of the American College of Radiology(ACR)Thyroid Imaging Reporting and Data System(TIRADS),the Chinese(C)-TIRADS,and the Artificial Intelligence(AI)-TIRADS for benign and malignant thyroid nodules,and to provide a basis for their clinical application.Methods A total of 102 thyroid nodules from 102 patients admitted to Hefei Second People's Hospital from April 2022 to May 2025 were included in the study.After pathological examination,diagnosed as malignant and included in the malignant group,and diagnosed as benign and included in the benign group.The clinical data,various ultrasound signs of the malignant and benign groups were compared.The malignant patients were regarded as positive cases,and the benign ones as negative controls.The ROC curve was used to analyze the diagnostic efficiency of ACR-TIRADS,C-TIRADS,and AI-TIRADS for benign and malignant thyroid nodules.Results Compared with the benign group,the malignant group had a higher proportion of solid internal composition,low/very low echo,irregular or expansive borders,and an aspect ratio≥1(P<0.05).A total of 71 cases of malignant thyroid nodules were diagnosed by pathology.ACR-TIRADS,C-TIRADS and AI-TIRADS correctly diagnosed 59 cases,61 cases and 65 cases of malignant thyroid nodules respectively.The results of ROC curve analysis showed that the area under the curve(AUC)of ACR-TIRADS,C-TIRADS and AI-TIRADS was 0.835(95%CI:0.756-0.914,P<0.05),0.865(95%CI:0.793-0.934,P<0.05),and 0.909(95%CI:0.847-0.972,P<0.05),with sensitivities of 83.10%,85.92%,91.55%and specificities of 83.87%,87.10%,90.32%,respectively.The DeLong test revealed that the AUC area of AI-TIRADS was significantly higher than that of ACR-TIRADS(P=0.008,95%CI:0.020-0.129)and C-TIRADS(P=0.037,95%CI:0.003-0.086),but there was no statistically significant difference in the AUC between ACR-TIRADS and C-TIRADS(P=0.110,95%CI:-0.007-0.067).After Holm multiple comparison correction,the difference in AUC between AI-TIRADS and ACR-TIRADS was still statistically significant(adjusted P=0.024),while the difference between AI-TIRADS and C-TIRADS was marginally significant after correction(adjusted P=0.074).There was no statistically significant difference between ACR-TIRADS and C-TIRADS(adjusted P=0.110).Conclusion For benign and malignant thyroid nodules,the diagnostic efficiency of AI-TIRADS was higher,C-TIRADS was had strong clinical operability and could optimize the diagnosis and treatment process of thyroid nodules.关键词
甲状腺结节/美国放射学会甲状腺影像报告与数据系统/中国甲状腺影像报告与数据系统/人工智能甲状腺影像报告与数据系统/受试者工作特征Key words
thyroid nodules/American College of Radiology Thyroid Imaging Reporting and Data System/Chinese Thyroid Imaging Reporting and Data System/Artificial Intelligence Thyroid Imaging Reporting and Data System/receiver operating characteristic引用本文复制引用
徐杰,魏杰,范伟健,马芳..ACR-TIRADS、C-TIRADS与AI-TIRADS在甲状腺结节良恶性诊断中的应用效果[J].分子影像学杂志,2026,49(5):604-610,7.基金项目
安徽省卫生健康科研项目(AHWJ2022c001) （AHWJ2022c001）
合肥市第二人民医院(安徽医科大学附属合肥医院)科研基金项目(2024ykc046) （安徽医科大学附属合肥医院）
