Machine Learning-based Single-cell Analysis Using Microfluidic Impedance Flow Cytometer

Single-cell analysis offers a more comprehensive approach to disease diagnosis compared to conventional methods. Electrical properties at a cellular level have been established as reliable biomarkers, enabling the identification of variations between individual cells. In this work we introduce a machine learning-based methodology for analyzing electrical impedance signals obtained from a microfluidic biosensor system for biological cell analysis. The proposed model is designed to detect and enumerate CD4 T-lymphocytes (CD4), which are a critical component of the immune system, through a microfluidic impedance flow cytometer. By identifying and analyzing the bioelectrical signal characteristics of CD4 cells as they traverse the sensing region, the machine learning models provide accurate cell enumeration while also estimating the size distribution of cell populations within the sample. A signal classification framework is employed to isolate cell signals from background noise, followed by the application and evaluation of various machine learning algorithms to optimize performance. The proposed method demonstrates improved accuracy and speed in cellular analysis compared to traditional techniques such as flow cytometry. Moreover, this method presents a significant potential for applications in cell analysis, addressing the demand for point-of-care diagnostics and enhancing the efficiency of biological diagnostics.