机构地区:[1]成都医学院第一附属医院放射科,四川成都610500 [2]成都医学院第一附属医院超声科,四川成都610500
出 处:《中国CT和MRI杂志》2022年第9期52-54,共3页Chinese Journal of CT and MRI
摘 要:目的 分析基于人工智能(AI)技术的肺部X线计算机断层摄影(CT)扫描在肺结节筛查及良恶性鉴别诊断中的应用价值。方法 回顾性分析2017年3月至2019年12月我院127例行肺部CT扫描的肺结节患者临床资料。记录经病理检查及临床判断的肺结节良恶性状况,分析基于AI技术CT扫描与专业医师CT扫描对良恶性肺结节的诊断价值;并由AI技术影像报告及2名专业医师确定真阳性结节,分别比较基于AI技术CT扫描与专业医师CT扫描对真阳性结节的检出情况。结果 127例患者中103倒(114个病灶)经病理检查确诊,24例(37个病灶)经临床治疗及随访确诊;基于AI技术CT扫描诊断良恶性肺结节的准确率为95.36%,Kappa值为0.905,完全一致;专业医师CT扫描诊断良恶性肺结节的准确率为96.03%,Kappa值为0.919,完全一致;AI技术及专业医师对良恶性肺结节均具有极高的诊断价值。127例患者共发现548个真阳性结节,基于AI技术CT扫描检出540个(98.54%),专业医师CT扫描检出255个(46.53%),基于AI技术CT扫描对真阳性结节的漏诊率明显低于专业医师(P<0.05),但误诊率明显高于专业医师(P<0.05)。专业医师CT扫描对不同直径真阳性结节检出率均低于基于AI技术CT扫描(P<0.05),对实性和磨玻璃真阳性结节检出率也低于基于AI技术CT扫描(P<0.05)。结论 基于AI技术CT扫描对肺结节筛查效果较好,虽然假阳性率偏高,但也能有效鉴别良恶性结节,临床应用价值较高。Objective To analyze the application value of lung X-ray computed tomography(CT) scan based on artificial intelligence(AI) technology in lung nodule screening and benign and malignant differential diagnosis.Methods The clinical data of 127 patients with lung nodules who underwent lung CT scan from March2017 to December 2019 in our hospital were retrospectively analyzed.The benign and malignant status of lung nodules after pathological examination and clinical judgment were recorded,and the diagnostic value of CT scan based on AI technology and CT scan by professional physicians on benign and malignant lu ng nodules was analyzed.The true positive nodules were determined by AI technology imaging report and two professionalphysicians,and the detection conditions of true positive nodules by CT scan based on AI technology and CT scan by professional physicians were compared.Results Among 127 patients,103 case(114 lesions) were diagnosed by pathological exa mination,and 24 cases(37 lesions) were diagnosed by clinical treatment and follow-up.The accuracy rate of CT scan based on AI technology was 95.36% in the diagnosis of benign and malignant lung nodules,and the Kappa value was 0.905,with complete consistency.The accuracy rate of CT scan by professional physicians was 96.03% in the diagnosis of benign and malignant lung nodules,and the Kappa value was 0.919,with complete consistency.AI technology and professional physicians had extremely high diagnostic value on benign and malignant lung nodules.A total of 548 truepositive nodules were found among127 patients,and 540 nodules(98.54%) were detected by CT scan based on AI technology and 255 nodules(46.53%) were detected by CT scan by professional physicians.The missed diagnosis rate CT scan based on AI technology was significantly lower than that of professional physicians(P<0.05),but the misdiagnosis rate was significantly higher than that of professional physicians(P<0.05).The detection rate of CT scan by professional physicians for true positive nodules of different d
分 类 号:R445.3[医药卫生—影像医学与核医学] R563[医药卫生—诊断学]
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...