基于人工智能的内镜诊断在早期食管鳞癌的筛查方法构建及应用研究  

Method construction and application research of artificial intelligence-based endoscopic diagnosis in screening of early esophageal squamous cell carcinoma

作　　者：黄亚彬[1] 李锋[1] 贾爱群[1] 薛全胜 孙海丽[1] 冯亚东 HUANG Yabin;LI Feng;JIA Aiqun;XUE Quansheng;SUN Haili;FENG Yadong(Department of Gastroenterology,the People’s Hospital of Rugao,Jiangsu,Rugao 226500,China;Department of Gastroenterology,Zhongda Hospital Affiliated to Southeast University,Jiangsu,Nanjing 210000,China)

机构地区：[1]江苏省如皋市人民医院消化内科,江苏如皋226500 [2]东南大学附属南京中大医院消化内科,江苏南京210000

出　　处：《中国医药科学》2023年第11期8-11,共4页China Medicine And Pharmacy

基　　金：江苏省南通市科技计划项目(MSZ21087)。

摘　　要：目的探讨与研究食管癌高危人员早期食管鳞癌基因筛查的方法构建与应用研究,同时与内镜检出率比较研究,探索适用于大量食管鳞癌高危人群的早发现、早诊断的筛查方案。方法采用回顾性研究方法,选择2020年6月至2022年6月在江苏如皋市人民医院进行诊治的食管病变患者380例。所有患者空腹口服“食康一号”胶囊,继后缓慢饮用温水300 ml,约3 min后,沿牵引线缓慢细胞收集器(海绵体),放入固定液中,送专门的细胞学基因检测,进行细胞基因检测。同时所有患者给予内镜检查,于病变可疑处取病理组织送检,如内镜无明确病灶的,分别于食管上段、中段、贲门处三处取组织送检。结果在380例患者中,病理诊断为食管鳞癌56例,非食管鳞癌324例。内镜诊断为食管鳞癌45例,其鉴别诊断食管鳞癌的敏感度与特异度分别为78.57%(44/56)和99.69%(323/324)。基于人工智能的内镜诊断为食管鳞癌55例,其鉴别诊断食管鳞癌的敏感度与特异度分别为98.21%(55/56)和100.00%(324/324)。基于人工智能的内镜鉴别诊断早期食管鳞癌的敏感度优于内镜。结论基于人工智能的早期食管鳞癌筛查方法构建及应用能提高诊断的敏感度,有利于食管鳞癌的早期发现。Objective To investigate and study the method construction and application research of screening of early esophageal squamous cell carcinoma in high-risk personnel of esophageal cancer,and to compare its detection rate with that of endoscopy simultaneously,so as to explore a screening scheme suitable for early detection and diagnosis of a large number of high-risk personnel of esophageal squamous cell carcinoma.Methods A total of 380 cases of patients with esophageal lesions diagnosed and treated in the People’s Hospital of Rugao in Jiangsu Province from June 2020 to June 2022 were selected according to the retrospective research method.All patients took"Shikang No.1"capsule orally on an empty stomach,and then slowly drank 300 ml warm water.After about 3 minutes,the cell collector(sponge)was slowly put along the traction line into a fixed solution,and it was sent for a special cytological genetic test to conduct cell genetic testing.Simultaneously,all patients were given endoscopic examination,and pathological tissues were taken for inspection at the suspected lesion.If there was no definite lesion in the endoscope,tissues at the upper,middle and cardia of esophagus were taken for inspection,respectively.Results In 380 patients,56 cases were diagnosed as esophageal squamous cell carcinoma and 324 cases as non-esophageal squamous cell carcinoma.A total of 45 cases of esophageal squamous cell carcinoma were diagnosed by endoscopy,with the sensitivity and specificity of differential diagnosis of esophageal squamous cell carcinoma of 78.57%(44/56)and 99.69%(323/324),respectively.A total of 55 cases of esophageal squamous cell carcinoma were diagnosed by endoscopy based on artificial intelligence,with the sensitivity and specificity of differential diagnosis of esophageal squamous cell carcinoma of 98.21%(55/56)and 100.00%(324/324),respectively.The sensitivity of endoscope based on artificial intelligence in differential diagnosis of early esophageal squamous cell carcinoma was superior to that of endoscope.Conclusi

关 键 词：食管癌 高危人员 人工智能 内镜 细胞收集器 诊断效果 

分 类 号：R445[医药卫生—影像医学与核医学]