A neural network approach for real-time particle/cell characterization in microfluidic impedance cytometry

Abstract

Microfluidic applications such as active particle sorting or selective enrichment require particle classification techniques that are capable of working in real time. In this paper, we explore the use of neural networks for fast label-free particle characterization during microfluidic impedance cytometry. A recurrent neural network is designed to process data from a novel impedance chip layout for enabling real-time multiparametric analysis of the measured impedance data streams. As demonstrated with both synthetic and experimental datasets, the trained network is able to characterize with good accuracy size, velocity, and crosssectional position of beads, red blood cells, and yeasts, with a unitary prediction time of 0.4 ms. The proposed approach can be extended to other device designs and cell types for electrical parameter extraction. This combination of microfluidic impedance cytometry and machine learning can serve as a stepping stone to real-time single-cell analysis and sorting.

Positional dependence of particles and cells in microfluidic electrical impedance flow cytometry: origin, challenges and opportunities

“…Most of the presented devices have shown the ability to deal with size heterogeneous samples. Beyond that, it is also possible to provide cell/particle position and accurate characterization at the same time 45,138,140,141 .…”

“…In 2019, this method was also used to measure the position of particles before and after DEP focusing 128 . More recently, Honrado et al 140 used the new wiring scheme in two consecutive facing pairs of electrodes to measure both the lateral position and the height. The implementation of artificial intelligence approaches instead of classical curve fitting helped to fasten dramatically the extraction of the size, cross-sectional position and velocity from the data obtained.…”

“…Neural networks have been implemented for cell characterization instead of classical template curve fitting. It fastened dramatically the feature extraction of the tested cells, allowing real-time characterization 140 .…”

Insulator Based Dielectrophoresis: Micro, Nano, and Molecular Scale Biological Applications

“…Once the machine learning algorithm builds predicted data of dielectric spectra, one can then backtrack the DEP device parameters. Similarly, Honrado et al [ 131 ] used machine learning to evaluate and record electrical fingerprints for the accurate detection of thousands of red blood cells and yeasts [ 131 ]. The work by Lannin and Honrado [ 128 , 130 , 131 ] shows that machine learning can be employed in the iDEP device as well, since the only integration with the device is the image based analysis and data management with the predictable machine learning algorithms.…”

“…Similarly, Honrado et al [ 131 ] used machine learning to evaluate and record electrical fingerprints for the accurate detection of thousands of red blood cells and yeasts [ 131 ]. The work by Lannin and Honrado [ 128 , 130 , 131 ] shows that machine learning can be employed in the iDEP device as well, since the only integration with the device is the image based analysis and data management with the predictable machine learning algorithms. We believe that in the near future, integrating machine learning with real-time data analysis in a DEP device will further enhance our understanding of fluid micro-environments, enabling the optimized manipulation of biological specimens and prediction of sample parameters, such as polarization rate of proteins and dielectric spectra of cells.…”