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

Abstract

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.

Keywords

• artificial intelligence
• liquid-based cytology
• dna quantitative analysis
• pleural and ascites effusion
• diagnostic value