Automation of flow cytometry data analysis with elastic image registration

Area of Science:

• Immunology
• Computational Biology
• Biotechnology

Background:

• Manual gating in flow cytometry is subjective, time-consuming, and prone to errors.
• Existing automated methods struggle with highly variable or continuously expressed markers.

Purpose of the Study:

• To develop and validate an automated gating tool, BD ElastiGate Software, that replicates manual gating processes.
• To improve the efficiency and accuracy of flow cytometry data analysis.

Main Methods:

• ElastiGate uses visual pattern recognition and elastic B-spline image registration to adjust gates based on local data variability.
• The tool converts flow cytometry plots into images for automated transformation and analysis.
• Validation was performed on diverse datasets including CAR-T cell manufacturing and tumor immunophenotyping.

Main Results:

• ElastiGate achieved performance comparable to manual gating, with average F1 scores exceeding 0.9 across all tested gates.
• The software demonstrated superior performance compared to existing 2D autogating solutions.
• Analysis time was substantially reduced across multiple applications.

Conclusions:

• BD ElastiGate Software provides an accurate and efficient automated solution for flow cytometry gating.
• The tool is accessible as a plugin for FlowJo Software and integrated into BD FACSuite Software.
• ElastiGate offers a robust alternative to manual gating, particularly for complex and variable datasets.