High-throughput cell spheroid production and assembly analysis by microfluidics and deep learning.

Area of Science:

• Biotechnology
• Cell Biology
• Artificial Intelligence

Background:

• 3D cell cultures are vital for translational research but limited in high-throughput screening.
• Challenges include complexity, high cell numbers, and poor standardization.

Purpose of the Study:

• To develop a high-throughput workflow for producing and characterizing miniaturized 3D cell culture spheroids.
• To leverage microfluidics and deep learning for scalable spheroid production and analysis.

Main Methods:

• Utilized droplet microfluidics for miniaturized spheroid production.
• Trained a convolutional neural network (CNN) for cell morphology classification.
• Optimized surfactant concentrations and incubation times for spheroid assembly.

Main Results:

• Successfully established a high-throughput workflow for spheroid production.
• Demonstrated CNN performance in morphology classification, outperforming conventional methods.
• Characterized optimal conditions for minispheroid production across three cell lines.

Conclusions:

• The developed workflow and CNN provide a template for large-scale minispheroid production and analysis.
• This method is compatible with high-throughput screening and drug discovery.
• The system can be adapted to study spheroid responses to various conditions and compounds.