Automated, high-throughput quantification of EGFP-expressing neutrophils in zebrafish by machine learning and a...

Area of Science:

• Immunology
• Developmental Biology
• Bio-imaging

Background:

• The zebrafish (Danio rerio) is a valuable model organism for studying immune responses due to conserved immune cell lineages and transparent larvae.
• Traditional methods for analyzing immune cells in zebrafish larvae are low-throughput and labor-intensive.
• High-throughput screening is crucial for genetic and chemical exposure studies.

Purpose of the Study:

• To develop and validate a rapid, high-resolution, automated method for counting fluorescently labeled immune cells in zebrafish larvae in vivo.
• To enable high-throughput genetic and chemical screens using zebrafish as a model.

Main Methods:

• Utilized a Multi-Camera Array Microscope (MCAM™) for parallelized, high-resolution imaging of 96-well plates of zebrafish larvae.
• Developed a machine learning segmentation algorithm for identifying the optimal focal plane of each larva.
• Employed pixel intensity thresholding and blob detection for counting fluorescent neutrophils (expressing EGFP) in vivo.

Main Results:

• Acquired 18 gigapixels of image data from a full 96-well plate in 75 seconds.
• Automated cell counting was completed within 5 minutes.
• Validated the algorithmic counts against manual counts, demonstrating accuracy and utility for screens measuring neutrophil number changes.

Conclusions:

• The MCAM™ combined with machine learning provides a rapid and efficient method for in vivo immune cell quantification in zebrafish.
• This approach significantly enhances throughput for genetic and chemical screening applications.
• An open-source software package is provided for custom model training and analysis, adaptable for other cell types and species.