AI超声辅助诊断系统在低年资医师诊断甲状腺结节良恶性中的应用  

The Application of AI Ultrasound-assisted Diagnosis System in Diagnosing Benign and Malignant Thyroid Nodules for Junior Sonographers

作　　者：麦武平 周美君 胡叶 冀将婷 黎金华 刘红梅 Mai Wuping;Zhou Meijun;Hu Ye;Ji Jiangting;Li Jinhua;Liu Hongmei(Department of Ultrasound,Institute of Ultrasound in Musculoskeletal Sports Medicine,Guangdong Second Provincial General Hospital,Guangzhou,Guangdong 510317,China)

机构地区：[1]广东省第二人民医院超声科广东省第二人民医院肌骨运动医学超声研究所,广州市510317

出　　处：《中国超声医学杂志》2021年第3期248-251,共4页Chinese Journal of Ultrasound in Medicine

摘　　要：目的探讨人工智能超声辅助诊断系统(AI超声系统)是否有助于提高低年资医师对于甲状腺结节良恶性的诊断能力及其在甲状腺癌筛查中的应用价值。方法以在我院行甲状腺手术及病理检查的256例患者,共506个甲状腺结节作为研究对象,两名低年资医师与AI超声系统分别对术前甲状腺结节超声图像进行良恶性分类;之后两名低年资医师分别联合AI超声系统的诊断结果进行良恶性判读的调整。以甲状腺结节手术病理为"金标准",采用灵敏度、特异度和AUC来评估诊断效能。结果对于甲状腺结节良恶性诊断的灵敏度及AUC,AI超声系统均高于两名低年资医师(P<0.05)。两名医师联合AI超声系统后,诊断的灵敏度及AUC均有所提高(P<0.05)。结论AI超声系统对甲状腺结节的良恶性诊断具有较高的灵敏度及准确度,联合AI超声系统后,低年资医师可以提高对甲状腺癌诊断的准确度。Objective To explore whether the artificial intelligence ultrasound-assisted diagnosis system(AI ultrasound system)can help improve the diagnostic ability of the junior sonographers in the diagnosis of benign and malignant thyroid nodules and its application value in the thyroid screening.Methods A total of 506 thyroid nodules from 256 patients who underwent thyroid surgery and pathological examination in our hospital were collected.Two junior sonographers and AI ultrasound system retrospectively determined the benign or malignant classification.After wards,the two junior sonographers respectively combined with the results of the AI ultrasound system and adjusted the benign and malignant classification.The postoperative pathological diagnosis of thyroid nodules was performed as the"gold standard".The sensitivity,specificity and AUC were used to evaluate the diagnostic efficacy.Results In the diagnosis of benign and malignant thyroid nodules,the sensitivity and AUC of AI ultrasound system were higher than that of the two junior sonographers(P<0.05),expectedly.The diagnostic sensitivity and AUC were improved after the two junior sonographers combining with the AI ultrasound system(P<0.05).Conclusions The AI ultrasound system has a high sensitivity and accuracy in the diagnosis of benign and malignant thyroid nodules.The accuracy of junior sonographers can be improved by combining with the AI ultrasound system.

关 键 词：甲状腺结节 人工智能 超声 低年资医师 

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