# In situ detection of dead cells from live cells via a DC plus low frequency AC resistive pulse sensor

## TL;DR

A new sensor distinguishes live and dead cells using a combination of AC and DC signals, achieving high accuracy without labeling.

## Contribution

A novel microfluidic sensor using combined AC and DC resistive pulse sensing for cell viability detection with 100% accuracy.

## Key findings

- Live and dead cells formed distinct clusters with maximum classification accuracy of 100%.
- The sensor achieved high accuracy for HUVECs treated with ethanol or staurosporine (STS).
- The approach reduces hardware and data processing complexity compared to existing methods.

## Abstract

Differentiation and detection of live and dead cells are critical for assessing cell viability in biomedical research, evaluating drug efficacy, and monitoring cytotoxicity in therapeutic applications. We present a microfluidic sensor that consists of two successive resistive pulse sensing channels. An excitation signal composed of a low-frequency AC (75 kHz) component and a DC bias was used to measure four key parameters. Through the AC measurement, differences in cell impedance causes variations in phase angle and voltage peak. From the DC measurement, cell size can be inferred from the resistive pulse magnitude, and the cell’s zeta potential is represented by the transit time difference. Human umbilical vein endothelial cells (HUVECs) and human mesenchymal stem cells (hMSCs) were used to demonstrate the device’s utility. A soft margin support vector machine (SVM) was applied to define the decision boundary based on analysis of the four parameters. For both cell types, live and dead cells formed distinct clusters, achieving maximum classification accuracies of up to 100%. Additionally, HUVECs treated with either ethanol or staurosporine (STS) were classified with accuracies up to 100%. Compared to previous microfluidic resistive pulse sensor (RPS), this approach can determine cell viability without the need for complex labeling or modifications. Unlike impedance cytometry, it does not require high-frequency measurements, significantly reducing hardware requirements and data processing complexity, while still providing multiparametric measurements of cells. These measurements allow the use of soft SVM to classify cell groups with higher accuracy than single-parameter differentiation.

The online version contains supplementary material available at 10.1007/s10544-026-00797-y.