论文部分内容阅读
目的:探讨胸水中腺癌细胞(ACC)和反应性间皮细胞(RMC)的计算机判别方法 ,为胸水中ACC计算机诊断奠定基础。方法:取胸水涂片中764个ACC(44例)和709个RMC(39例),用计算机图像分析技术分别测试细胞的色度学和形态结构参数,随机选取约80%细胞作为训练组,剩余的20%细胞为测试组,对训练组的测试结果进行逐步判别分析并建立判别函数,用测试组评价函数的判别效果。结果:逐步判别分析显示,RP、BP、Rn、Bn、dmaj,c、dmin,c、An、Dn、Cn进入判别函数,符合率为84.5%,其判别效果明显优于单独基于色度学和形态结构参数所建立的判别函数(P<0.05)。结论:结合色度学和形态结构参数所建立的判别函数判别效果好,为胸水中腺癌细胞的计算机定量诊断提供了分析路径。
Objective: To investigate the computer-based discriminant method of pleural effusion in adenocarcinoma cells (ACC) and reactive mesothelial cells (RMCs), and lay a foundation for the computerized diagnosis of ACC in pleural effusion. Methods: 764 ACC (44 cases) and 709 RMC (39 cases) in pleural effusion smears were obtained. The chromatographic and morphological parameters of the cells were tested by computer image analysis. Approximately 80% of the cells were randomly selected as the training group. The remaining 20% of the cells were test groups. The discriminant function was established by stepwise discriminant analysis of the test results in the training group and the discriminant effect of the test function was evaluated by the test group. Results: Stepwise discriminant analysis showed that the discriminant functions of RP, BP, Rn, Bn, dmaj, c, dmin, c, An, Dn and Cn entered the discriminant function with a coincidence rate of 84.5% Discriminant function established by morphological parameters (P <0.05). CONCLUSION: The discriminant function based on the combination of colorimetric and morphological parameters has a good discriminating effect and provides an analytical path for the quantitative diagnosis of adenocarcinoma cells in pleural effusion.