Automated detection of cytokinesis-blocked micronuclei using fuzzy c-means algorithm and morphological features

Increased frequency of micronuclei is positively correlated with the molecular dosimetry of genotoxic damages. The cytokinesis-block micronucleus test (CBMN test) is a well-established assay used in toxicological screening for potential genotoxic compounds. Since the method is simple and economical, CBMN assay can be employed on a large scale as a quantitative biological dosimeter. Automated detection of binucleated cells and micronuclei helps to increase the speed and reproducibility of CBMN method and is very beneficial for large-scale biomonitoring applications. Fuzzy c-means algorithm as well as mathematical morphology was employed in this preliminary study to identify the interested structures and helped to develop a robust comprehensive program to analyze the CBMN slides automatically. As a result, all the three kinds of binucleated cells with different nuclear morphologies were identified by our program. The structures that did not satisfied the criteria of binucleated cells and micronuclei were successfully excluded by the program.