Automation of the Micronucleus Assay Using Imaging Flow Cytometry and Artificial Intelligence

Area of Science:

• Toxicology and Pharmacology
• Genetics and Genomics
• Computational Biology

Background:

• The micronucleus (MN) assay is a regulatory standard for assessing chemical genotoxicity.
• Traditional light microscopy for MN scoring is time-consuming and subjective.
• Flow cytometry offers high throughput but lacks visual confirmation capabilities.

Purpose of the Study:

• To describe the development and application of an AI-driven deep learning model for automated MN assay scoring.
• To integrate AI with imaging flow cytometry (IFC) for enhanced MN assay analysis.
• To validate AI-based scoring against manual microscopy for genetic toxicity assessment.

Main Methods:

• Utilized artificial intelligence (AI) and convolutional neural networks to create a deep learning model.
• Employed imaging flow cytometry (IFC) for high-throughput image acquisition and automated analysis of MN assay data.
• Developed a workflow for training AI models and applying them to score MN assay events.

Main Results:

• The AI deep learning model successfully scored all key events in the MN assay using IFC data.
• Results generated by the AI model demonstrated strong agreement with manual microscopy scoring.
• The combined IFC and AI approach enables fully automated scoring of the MN assay.

Conclusions:

• AI, particularly deep learning models, can automate the scoring of MN assay data acquired by IFC.
• This integrated approach provides a robust, efficient, and objective method for genotoxicity testing.
• Combining IFC and AI represents a significant advancement in the automated evaluation of genetic toxicity.