人工智能辅助甲状腺影像报告和数据系统4a类结节穿刺活检  

Needle Biopsy for Thyroid Imaging Reporting and Data System 4a Nodules Assisted by Artificial Intelligence

作　　者：何子朋[1] 郭瑞君[2] 唐华[1] HE Zipeng;GUO Ruijun;TANG Hua(Department of Ultrasound,West Campus,Beijing Chaoyang Hospital Affiliated to Capital Medical University,Beijing 100043,China;不详)

机构地区：[1]首都医科大学附属北京朝阳医院西院超声科,北京100043 [2]首都医科大学附属北京朝阳医院超声医学科,北京100020

出　　处：《中国医学影像学杂志》2021年第12期1195-1198,共4页Chinese Journal of Medical Imaging

摘　　要：目的探讨人工智能(S-Detect技术)辅助甲状腺影像报告和数据系统(TI-RADS)4a类结节穿刺活检的应用价值。资料与方法选取常规超声TI-RADS分类为4a类的甲状腺结节79例共86个病灶。应用S-Detect技术对所有病灶进行良恶性诊断后,再次进行分类。以病理结果作为“金标准”,比较S-Detect技术辅助分类前后的差异。结果S-Detect技术辅助诊断前,遵循临床治疗策略均应进行穿刺活检,活检率为100.0%(86/86)、特异度为0%(0/52)、敏感度为100.0%(34/34)、阳性预测值为39.5%(34/86)、误诊率为100.0%(52/52)、漏诊率为0%(0/34);S-Detect技术辅助诊断后,38个病灶由4a类降为3类,活检率为55.8%(48/86)、特异度为71.2%(37/52)、敏感度为97.1%(33/34)、阳性预测值为68.8%(33/48)、误诊率为28.8%(15/52)、漏诊率为2.9%(1/34),其中活检率和误诊率显著降低(χ^(2)=48.776、57.433,P均<0.01),特异度和阳性预测值显著提高(χ^(2)=57.433、10.517,P均<0.01),敏感度和漏诊率无明显变化(P>0.05)。结论S-Detect技术辅助甲状腺结节再次分类有助于提高TI-RADS 4a类结节的活检效能,降低穿刺率。Purpose To explore the application value of needle biopsy for thyroid imaging reporting and data system(TI-RADS)4a nodules assisted by artificial intelligence(S-Detect).Materials and Methods A total of 79 patients with 86 thyroid nodules classified as TIRADS 4a by conventional ultrasound(US)were enrolled.All the lesions were detected by S-Detect assisted US and classified again.The pathological results were regarded as gold standard.The differences between conventional US results and S-Detect assisted US results were compared.Results Before S-Detect application,as all lesions were proposed for biopsy follow the clinical treatment stategy,the biopsy rate,specificity,sensitivity,positive predictive value,misdiagnosis rate and missed diagnosis rate were 100.0%(86/86),0%(0/52),100.0%(34/34),39.5%(34/86),100.0%(52/52),0%(0/34),respectively.After reclassification with S-Detect,38 lesions were downgraded to TI-RADS category 3.The biopsy rate,specificity,sensitivity,positive predictive value,misdiagnosis rate and missed diagnosis rate were 55.8%(48/86),71.2%(37/52),97.1%(33/34),68.8%(33/48),28.8%(15/52),2.9%(1/34),respectively.With the assist of S-Detect,the biopsy rate and misdiagnosis rate reduced significantly(χ^(2)=48.776,57.433,P<0.01),the specificity and positive predictive value increased significiantly(χ^(2)=57.433,10.517,P<0.01),while the sensitivity and missed diagnosis rate changed with no statistical difference(P>0.05).Conclusion Using S-Detect as a reference criteria for thyroid nodule reclassification increased the biopsy efficiency of nodules classified as TI-RADS 4a and reduced the puncture rate.

关 键 词：甲状腺结节 超声检查 甲状腺影像报告和数据系统 S-Detect 病理学 外科 诊断 鉴别 

分 类 号：R445.1[医药卫生—影像医学与核医学] R581.3[医药卫生—诊断学]