Using artificial intelligence to improve cell therapy assays: automated quantitative image analysis of cells on...

Area of Science:

• Cell Therapy
• Biotechnology
• Image Analysis

Background:

• Cell therapy advancements utilize directed delivery methods like 3D scaffolds.
• Accurate cell counting and viability assessment are crucial for process optimization.
• Current manual methods for cell quantification are labor-intensive and have limited dynamic range.

Purpose of the Study:

• To develop and validate a simple, automated, and quantitative fluorescent imaging-based method for counting live and dead cells in scaffold cultures.
• To compare the performance of traditional image quantitation software with a trained artificial intelligence (AI) software for cell counting and viability analysis.
• To assess the suitability of AI-based approaches for enhancing cell therapy development.

Main Methods:

• Optimized labware for uniform imaging fields in scaffold cultures.
• Compared traditional image quantitation software (Gen5) with trained AI software (Aiforia) for cell counting and viability.
• Utilized fluorescent imaging with Propidium Iodide for dead cell detection.

Main Results:

• AI software (Aiforia) showed high correlation between live cell counts and seeding concentration (r²=0.96).
• Traditional software (Gen5) showed no correlation for live cell counts (r²=0.09).
• Both methods showed correlation for dead cell counts (r²=0.90), indicating comparable detection.

Conclusions:

• Trained AI software significantly improves data accuracy for cell counting in complex scaffold backgrounds.
• AI-based image analysis offers a more consistent and quantitative solution for cell therapy applications.
• This automated method supports process optimization and release testing in cell therapy development.