Supervised machine learning in microfluidic impedance flow cytometry for improved particle size determination.

Area of Science:

• Biomedical Engineering
• Microfluidics
• Cell Biology

Background:

• Accurate particle and cell size assessment is crucial in electrical microfluidic flow cytometry.
• Coplanar electrodes in flow cytometers create inhomogeneous electric fields, complicating precise particle size determination.
• Existing methods using signal templates and compensation are cumbersome for irregular signal shapes.

Purpose of the Study:

• To develop a simple, accurate post-processing method for particle and cell size determination in microfluidic flow cytometry.
• To overcome limitations of current methods in handling inhomogeneous electric fields and irregular signal shapes.
• To improve the precision of cell size analysis in impedance-based flow cytometers.

Main Methods:

• Implementation of a supervised machine learning approach, specifically a multiple linear regression model.
• Development of a post-processing technique that avoids pre-defined signal templates and compensation functions.
• Application of the method to determine the size distribution of yeast cell populations.

Main Results:

• Achieved an average reduction of 37% in particle diameter variation compared to previous feature extraction and compensation methods.
• Successfully demonstrated the method's efficacy in determining the size distribution of small (4.6 ± 0.9 μm) and large (5.9 ± 0.8 μm) yeast cells.
• Validated the improved performance of a coplanar, two-electrode chip for precise cell size determination.

Conclusions:

• The developed machine learning method offers a simple and accurate alternative for cell size assessment in microfluidic flow cytometry.
• This approach effectively addresses the challenges posed by inhomogeneous electric fields in coplanar electrode systems.
• The enhanced precision enables reliable cell size determination in easily fabricated impedance flow cytometers.