Machine Learning-Enhanced Microfluidic Impedance Platform for Rare Cell Analysis

Area of Science:

• Biomedical Engineering
• Cell Biology
• Analytical Chemistry

Background:

• Rare cells, including cancer cells, significantly impact health and disease but are challenging to detect due to low abundance.
• High-precision methods are crucial for identifying and analyzing these rare cell populations.
• Microfluidic impedance-based flow cytometry (μIFC) offers a label-free approach for single-cell analysis.

Purpose of the Study:

• To develop and validate a multifrequency μIFC platform integrated with a support vector machine (SVM) learning strategy.
• To accurately distinguish rare cancer cells from white blood cells (WBCs) using a label-free method.
• To enhance the reliability of rare cell detection through a post-prediction correction strategy.

Main Methods:

• Utilized a multifrequency microfluidic impedance-based flow cytometry (μIFC) platform.
• Implemented a support vector machine (SVM) learning strategy for cell classification.
• Applied a post-prediction correction strategy to reduce false identification rates.
• Validated performance by differentiating cancer cell lines (MDA-MB-231, A549, HeLa) from lymphocytes (Jurkat) and PBMCs.

Main Results:

• Achieved over 99% accuracy in differentiating cancer cell lines from WBCs.
• Demonstrated the platform's ability to detect rare cancer cells within WBC populations at 1-10% concentrations.
• Showed high consistency between μIFC results and conventional flow cytometry.
• Reduced false identification rates using the post-prediction correction strategy.

Conclusions:

• Established a robust, label-free, machine learning-enhanced μIFC platform for rare cancer cell analysis.
• The developed platform offers high accuracy and reliability for distinguishing rare cancer cells from WBCs.
• This technology holds promise for advancing rare cell characterization and related applications.