High-content video flow cytometry with digital cell filtering for label-free cell classification by machine learning.

Area of Science:

• Biomedical Engineering
• Cell Biology
• Machine Learning

Background:

• Imaging flow cytometry (IFC) enables high-throughput single-cell analysis using fluorescent labels.
• Fluorescent labels can interfere with cell function and limit image quality in high-throughput settings.
• Limitations exist in current methods for high-throughput, label-free single-cell analysis.

Purpose of the Study:

• To develop a high-content video flow cytometry (VFC) system for label-free, high-throughput single-cell analysis.
• To implement machine learning for automatic digital cell filtering and classification.
• To assess the VFC system's performance in differentiating cervical carcinoma cell lines.

Main Methods:

• Development of a high-content video flow cytometry (VFC) system capable of measuring unlabeled single cells at ~1000 cells/minute.
• Application of a digital cell filtering technique using machine learning to process large datasets and identify the frame of interest (FOI).
• Utilizing deep learning for three-way classification of cervical carcinoma cell lines (Caski, HeLa, C33-A).

Main Results:

• The VFC system provides high-quality, high-throughput, and rewinding images of single cells without fluorescent labels.
• Machine learning-based digital cell filtering effectively processed big data.
• Deep learning achieved high classification accuracies: 91.5% for Caski, 90.5% for HeLa, and 90.5% for C33-A cells.

Conclusions:

• High-content VFC offers a label-free approach for high-throughput single-cell imaging and analysis.
• The developed system demonstrates potential for automatic digital cell filtering and accurate cell classification.
• This label-free VFC technology may have significant clinical applications in cell diagnostics.