Electrical micro flow cytometry with LSTM and its application in leukocyte differential

Area of Science:

• Biomedical Engineering
• Cellular Biology
• Electrical Engineering

Background:

• Leukocyte differential counts are crucial for diagnosing various diseases.
• Traditional flow cytometry methods can be complex and expensive.
• There is a need for more accessible and efficient cell analysis techniques.

Purpose of the Study:

• To develop an electrical micro flow cytometry system for accurate leukocyte differential analysis.
• To leverage a constrictional microchannel and a deep neural network (RNN) for cell classification.
• To validate the system's performance with both purified and mixed leukocyte samples.

Main Methods:

• Cells flowed through a constrictional microchannel, generating impedance variations.
• A recurrent neural network (RNN) was optimized for analyzing electrical signals.
• The trained RNN classified impedance profiles from mixed leukocyte samples.

Main Results:

• The optimized RNN achieved 95.2% classification accuracy for purified leukocytes.
• Analysis of mixed leukocyte samples showed consistent differential results (e.g., granulocytes 57% vs. 59%).
• The system demonstrated reliable performance in differentiating white blood cell types.

Conclusions:

• The developed electrical micro flow cytometry system effectively differentiates leukocytes.
• The combination of a constrictional microchannel and RNN offers a novel approach for cell analysis.
• This technology provides a new perspective for electrical flow cytometry in white blood cell differentiation.