AI细胞形态学联合DNA定量分析鉴别良恶性胸腹水的探讨

AI Cytomorphology Combined with DNA-image Cytometry for Identifying Benign and Malignant Pleural Effusion and Ascites

目的探讨通过人工智能(AI)细胞学联合DNA定量分析(DNA-ICM)辅助诊断系统在良恶性胸腹水鉴别中的诊断价值。方法用液基细胞学(LCT)、DNA-ICM、AI、AI联合DNA-ICM系统对360例胸腹水标本进行良恶性鉴别,比较几种检测方法的敏感度、特异性、准确度、Kappa值、约登指数及曲线下面积。结果通过AI联合DNA-ICM检测良恶性胸腹水的敏感度、特异性、准确度分别为95.23%、94.12%、94.44%,高于其他三种单独检测方法,差异均具有统计学意义(P<0.05)。LCT、DNA-ICM、AI检测的Kappa值分别为0.646、0.642、0.586,约登指数分别为0.693、0.687、0.676,AUC分别为0.846、0.843、0.838;通过AI联合DNA-ICM的Kappa值为0.869,约登指数为0.893,AUC为0.947,均高于三种单独检测方法。结论三种单独检测方法中LCT的可靠性、真实性、诊断价值最高,可为临床鉴别良恶性胸腹水的常用方法。通过AI联合DNA-ICM辅助诊断系统阅片,在鉴别良恶性胸腹水的诊断效能比三种单独检测方法好,可作为临床鉴别良恶性胸腹水的可靠方法。

Objective To explore the diagnostic value of artificial intelligence(AI)cytology combined with DNA-image cytometry(DNA-ICM)auxiliary diagnostic system for the identification of benign and malignant pleural effusion and ascites.Methods Liquid-based cytology technology(LCT),DNA-ICM,AI,and AI combined with DNA-ICM were used to identify benign and malignant pleural effusion and ascites specimens in 360 cases,and their sensitivity,specificity,accuracy,Kappa value,Youden index and AUC were statistically analyzed.Results The sensitivity,specificity,and accuracy of AI combined with DNA-ICM in detecting benign and malignant pleural effusion and ascites were 95.23%,94.12%,and 94.44%,respectively,which were higher than those of the three other separate detection methods(all P<0.05).The kappa values of LCT,DNA-ICM,and AI were 0.646,0.642,and 0.586;their Youden index values were 0.693,0.687,and 0.676,and their AUC values were 0.846,0.843,and 0.838,respectively.The Kappa value of AI combined with DNA-ICM was 0.869,the Youden index was 0.893,and AUC was 0.947,which were all higher than those of the three detection methods alone.Conclusion Among the three separate detection methods,LCT has the highest reliability,authenticity,and diagnostic value,and it can be used as a common method for the clinical identification of benign and malignant pleural effusion and ascites.The diagnostic performance of AI combined with DNA-ICM auxiliary diagnosis system in identifying benign and malignant pleural effusion and ascites is better than those of the three separate detection methods and can be used as a reliable method for the clinical identification of benign and malignant pleural effusion and ascites.