人工智能技术与FNAC联合在甲状腺良恶性结节诊断中的应用价值分析  

Application Value of Artificial Intelligence Technology Combined with FNAC in the Diagnosis of Benign and Malignant Thyroid Nodules

作　　者：葛肖艳[1] 马婷 刘文[1] GE Xiaoyan;MA Ting;LIU Wen(Department of Ultrasound Medicine,First Affliated Hospital of Shihezi University)

机构地区：[1]石河子大学第一附属医院超声医学科

出　　处：《中国医学计算机成像杂志》2024年第5期589-593,共5页Chinese Computed Medical Imaging

基　　金：兵团项目(2023ZD004);石河子大学科研项目(ZZZC2022072)。

摘　　要：目的:探究人工智能技术联合细针吸取细胞学检查(FNAC)在甲状腺良恶性结节诊断中的应用价值。方法:选取2022年1月至2023年12月在我院进行诊治的128例甲状腺结节患者作为研究对象并记录所有患者的临床资料,采用多普勒超声诊断仪分别进行甲状腺超声人工智能检查、超声引导下加负压FNAC以及二者联合检查,完成后统计患者的病理学结果,人工智能检测将内置算法作为指导,观察病灶参数,对良性、恶性例数进行统计;分析对比单一检测与联合检测的检出率、准确度、灵敏度、特异度、阳性及阴性预测值,并进行受试者工作特征(ROC)曲线分析。结果:128例甲状腺结节患者中,病理学检测出有良性结节者44例(34.38%),恶性结节者84例(65.63%);FNAC检测出良性结节患者55例,恶性结节患者73例;人工智能检测出良性结节患者49例,恶性结节患者79例;联合检测出良性结节患者45例,恶性结节患者83例。在各检测方式阳性与阴性预测值结果中,FNAC检测阳性预测值为65.45%,阴性预测值为89.04%;人工智能检测阳性预测值为71.42%,阴性预测值为88.61%;联合检测阳性预测值为95.55%,阴性预测值为98.80%。采用预测值对识别诊断绘制ROC发现,FNAC检测AUC值为0.789、灵敏度81.82%、特异度78.91%、约登指数0.60;人工智能检测AUC值为0.784、灵敏度79.55%、特异度79.54%、约登指数0.59;联合检测的AUC值为0.985、灵敏度97.72%、特异度92.28%、约登指数0.90,提示在甲状腺良恶性结节诊断中,与单一检测相比,联合检测具有更高的诊断价值。结论:与单一检测相比,人工智能技术与FNAC相结合的检测方式在甲状腺良恶性结节诊断中的诊断效率更高,建议临床推广使用。Purpose:To explore the application value of artificial intelligence technology combined with fine-needle aspiration cytology(FNAC)in the diagnosis of benign and malignant thyroid nodules.Methods:A total of 128 patients with thyroid nodules diagnosed and treated in our hospital from January 2022 to December 2023 were selected as the study subjects,and the clinical data of all patients were recorded.Ultrasonic artificial intelligence examination of thyroid,ultrasus-guided negative pressure FNAC and combined examination were performed using Doppler ultrasound diagnostic instrument.The pathological results of the patients were counted,and the artificial intelligence detection took the built-in algorithm as a guide to observe the lesion parameters.Benign and malignant cases were counted.The detection rate,accuracy,sensitivity,specificity,positive and negative predictive values of single detection and combined detection were compared,and receiver operating characteristic(ROC)curve analysis was performed.Results:Of the 128 patients with thyroid nodules,44(34.38%)were with benign nodules and 84(65.63%)with malignant nodules detected by pathology.FNAC detected 55 cases of benign nodules and 73 cases of malignant nodules.The artificial intelligence detected 49 cases of benign nodules and 79 cases of malignant nodules.Benign nodules were detected in 45 cases and malignant nodules in 83 cases by combined method.Among the positive and negative predictive values,the positive predictive value of FNAC was 65.45%and the negative predictive value was 89.04%.The positive predictive value of artificial intelligence was 71.42%and the negative predictive value was 88.61%.The positive predictive value of combined method was 95.55%and the negative predictive value was 98.80%.It was found that the area under ROC curve(AUC)value for FNAC detection was 0.789,with the sensitivity of 81.82%,specificity of 78.91%,and Jorden index of 0.60.The AUC value for artificial intelligence was 0.784,with the sensitivity of 79.55%,specificity of 79.54%,a

关 键 词：甲状腺 人工智能技术 细针吸取细胞学检查 诊断 应用价值 

分 类 号：R445.1[医药卫生—影像医学与核医学]