Automatic detection of micronuclei by cell microscopic image processing

Area of Science:

• Cytogenetics
• Radiation Biology
• Biomonitoring

Background:

• Ionizing radiation exposure necessitates monitoring its health effects, particularly cytogenetic damage like micronuclei (MNi).
• The cytokinesis-block micronucleus (CBMN) assay is standard for assessing genome instability, but manual scoring is laborious and prone to inconsistency.
• Limitations in current methods highlight the need for automated, reliable scoring solutions.

Purpose of the Study:

• To develop and validate automated software for scoring MNi in human lymphocytes.
• To overcome limitations of manual scoring in the CBMN assay, enhancing efficiency and consistency.
• To provide a reliable tool for biomonitoring radiation effects on human health.

Main Methods:

• Developed a four-stage automated scoring system: nuclei segmentation, binucleated (BN) cell detection, cell estimation, and MNi detection.
• The software is designed to handle BN cells with indistinct cytoplasm and MNi in peripheral blood smears.
• Tested the system on a custom dataset to evaluate performance.

Main Results:

• Achieved high sensitivity: approximately 98% for BN cell recognition and 82% for MNi detection.
• Demonstrated high reliability with less than 1% false positives.
• The software effectively identifies BN cells and MNi, even in challenging samples.

Conclusions:

• The developed software offers a reliable and efficient automated solution for MNi scoring in the CBMN assay.
• This tool can significantly improve biomonitoring of radiation exposure and assessment of genome instability.
• The high accuracy and low false positive rate support its practical application in cytogenetic analysis.