Upgraded User-Friendly Image-Activated Microfluidic Cell Sorter Using an Optimized and Fast Deep Learning Algorithm

Area of Science:

• Biotechnology
• Microfluidics
• Cell Biology

Background:

• Image-based cell sorting is crucial for biological and biomedical research, enabling downstream analysis of cell-to-cell variations.
• Previous work established a user-friendly image-activated microfluidic cell sorting technique leveraging a fast deep learning algorithm for real-time cell isolation.
• The technique was initially demonstrated using an inverted microscope setup.

Purpose of the Study:

• To implement and upgrade the previously demonstrated image-activated microfluidic cell sorting technique into a practical, real-world system.
• To incorporate new features enhancing user-friendliness and efficiency for real-time cell or particle sorting.
• To validate the performance of the upgraded sorting system.

Main Methods:

• Development of an upgraded cell sorting system integrating microscope-based techniques into a functional device.
• Incorporation of a high-resolution linear piezo-stage for in-focus imaging of rapidly moving cells.
• Integration of an LED strobe light to reduce motion blur and a vertical syringe pump to prevent cell sedimentation.

Main Results:

• The upgraded system demonstrated successful real-time sorting of fluorescent polystyrene beads.
• Achieved a high sorting purity of 99.4% for both 15 μm and 10 μm beads.
• Attained an average throughput of 22.1 events per second (eps).

Conclusions:

• The upgraded image-activated microfluidic cell sorting system offers enhanced performance and user-friendliness.
• The system's high purity and throughput make it a valuable tool for real-time cell and particle sorting in research.
• This advancement facilitates more detailed downstream analysis and expands knowledge of cellular differences.