期刊文献+

人工智能超声辅助诊断系统联合血清miR-211鉴别诊断甲状腺结节的临床价值

Clinical value of artificial intelligence ultrasound-assisted diagnostic system combined with serum miR-211 in the differential diagnosis of thyroid nodules
下载PDF
导出
摘要 目的探讨人工智能超声辅助诊断系统(以下简称AI超声系统)联合血清miR-211鉴别诊断良恶性甲状腺结节的临床应用价值。方法选取我院经病理证实的甲状腺结节患者203例(良性146例,恶性57例,均为单发),术前均先行AI超声系统检查评估结节良恶性,再分别由1名高年资超声医师和1名低年资超声医师评估结节良恶性;检测并比较良恶性甲状腺结节血清miR-211的差异。分析AI超声系统及不同年资超声医师诊断结果与病理结果的一致性,计算并比较其诊断效能。绘制受试者工作特征(ROC)曲线分析血清miR-211鉴别良恶性甲状腺结节的诊断效能,比较AI超声系统联合血清miR-211与其单独应用的诊断效能。结果AI超声系统诊断良性结节119例,恶性结节84例,与病理结果的一致性一般(Kappa=0.627,P<0.01),其准确率、灵敏度、特异度、阳性预测值、阴性预测值分别为82.76%、92.98%、78.77%、63.10%、96.64%;高年资超声医师诊断恶性结节124例,恶性结节79例,与病理结果的一致性一般(Kappa=0.607,P<0.01),其准确率、灵敏度、特异度、阳性预测值、阴性预测值分别为82.27%、87.72%、80.13%、63.29%、94.35%;低年资超声医师诊断良性结节105例,恶性结节98例,与病理结果的一致性较差(Kappa=0.470,P<0.01),其准确率、灵敏度、特异度、阳性预测值、阴性预测值分别73.89%、89.47%、67.81%、52.04%、94.29%。三者诊断准确率、特异度比较差异均有统计学意义(均P<0.05);其中AI超声系统和高年资超声医师诊断准确率、特异度均高于低年资超声医师(均P<0.05)。恶性结节血清miR-211高于良性结节,差异有统计学意义(Z=7.807,P<0.05)。ROC曲线分析显示,血清miR-211鉴别诊断良恶性甲状腺结节的曲线下面积为0.853,诊断准确率、灵敏度、特异度、阳性预测值、阴性预测值分别为84.24%、89.47%、82.19%、66.23%、95.24%。AI超声系统联合血清miR-211鉴别诊断良恶性甲状腺结节的准确率、灵敏度、特异度、阳性预测值、阴性预测值分别为90.15%、89.47%、90.41%、78.46%、95.65%,联合应用的诊断准确率、特异度、阳性预测值均高于AI超声系统和血清miR-211单独应用,差异均有统计学意义(均P<0.05)。结论AI超声系统在良恶性甲状腺结节鉴别诊断方面具有较好价值,其与miR-211联合应用可提高诊断准确率、特异度及阳性预测值。 Objective To investigate the clinical application value of artificial intelligence ultrasound assisted diagnosis system(hereinafter referred to as AI ultrasound system)combined with serum miR-211 in the differential diagnosis of benign and malignant thyroid nodules.Methods A total of 203 patients with pathologically confirmed thyroid nodules admitted to our hospital(146 benign and 57 malignant,all single)who underwent preoperative examination with the AI ultrasound system to determine the benign and malignant the nodules were selected.Then 1 senior sonographer and 1 junior sonographer were evaluated for benign and malignant nodules,respectively.The difference of serum miR-211 between with benign and malignant thyroid nodules were compared.The consistency of the AI ultrasound system and sonographers with different seniority in the diagnosis of benign and malignant thyroid nodules with pathological findings were analyzed,and their diagnostic efficacy were calculated and compared.Receiver operating characteristics(ROC)curve was drawn to analyzed the efficacy of serum miR-211 in the differential diagnosis of benign and malignant thyroid nodules,and the diagnostic efficacy of AI ultrasound system combined with serum miR-211 was compared with that applied alone.Results There were 119 benign and 84 malignant thyroid nodules diagnosed by AI ultrasound system,which was generally consistent with pathological results(Kappa=0.627,P<0.01),and the accuracy,sensitivity,specificity,positive predictive value,and negative predictive value were 82.76%,92.98%,78.77%,63.10%,and 96.64%,respectively.There were 124 benign and 79 malignant thyroid nodules diagnosed by senior sonographer,which was generally consistent with pathological results(Kappa=0.607,P<0.01),and the accuracy,sensitivity,specificity,positive predictive value,and negative predictive value were 82.27%,87.72%,80.13%,63.29%,and 94.35%,respectively.There were 105 benign and 98 malignant thyroid nodules diagnosed by junior sonographer,which was poor consistency with pathological results(Kappa=0.470,P<0.01),and the accuracy,sensitivity,specificity,positive predictive value,and negative predictive value were 73.89%,89.47%,67.81%,52.04%,and 94.29%,respectively.The differences in diagnostic accuracy and specificity among the three were statistically significant(all P<0.05).The diagnostic accuracy and specificity of AI ultrasound system and senior sonographer were higher than those of junior sonographer,and the differences were statistically significant(all P<0.05).The serum miR-211 of malignant nodules was higher than that of benign nodules,and the difference was statistically significant(Z=7.807,P<0.05).ROC curve analysis showed that the area under curve(AUC)of serum miR-211 in the differential diagnosis of benign and malignant thyroid nodules was 0.853,and the diagnostic accuracy,sensitivity,specificity,positive predictive value,and negative predictive value were 84.24%,89.47%,82.19%,66.23%,and 95.24%,respectively.The accuracy,sensitivity,specificity,positive predictive value and negative predictive value of AI ultrasound system combined with serum miR-211 in the differential diagnosis of benign and malignant thyroid nodules were 90.15%,89.47%,90.41%,78.46%and 95.65%,respectively.The diagnostic accuracy,specificity and positive predictive value of the combined application were higher than those of AI ultrasound system and serum miR-211 applied alone,and the differences were statistically significant(all P<0.05).Conclusion AI ultrasound system has a good value in the differential diagnosis of benign and malignant thyroid nodules,and its combined application with serum miR-211 can improve the diagnostic accuracy,specificity and positive predictive value.
作者 李莉 向丽丽 庹菲 刘芳 LI Li;XIANG Lili;TUO Fei;LIU Fang(Department of Ultrasound Imaging,the First Affiliated Hospital of Hunan University of Chinese Medicine,Changsha 410007,China)
出处 《临床超声医学杂志》 CSCD 2023年第12期1001-1005,共5页 Journal of Clinical Ultrasound in Medicine
关键词 超声辅助诊断系统 人工智能 血清miR-211 甲状腺结节 良恶性 Ultrasound-assisted diagnostic system,artificial intelligence Serum miR-211 Thyroid nodules,benign and malignant
  • 相关文献

参考文献6

二级参考文献32

共引文献230

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部