A Micirofluidic Flow Cytometry Enabling High-Throughput Characterization of Single-Cell Impedance and Imaging Based on Constrictional Microchannels Coupled with Deep Neural Networks

This paper presented a microfluidic flow cytometry enabling high-throughput captures of single-cell impedance and imaging based on constrictional microchannels coupled with deep neural networks. Based on constrictional microchannels, large impedance variations of single cells were obtained by effectively blocking the compressed electric lines and fluorescent images of high imaging qualities were captured by confining the travelling cells in the constrictional microchannels. Hundreds of single-cell features were extracted from impedance profiles and fluorescent images by a recurrent neural network and a convolutional neural network, respectively. Single-cell features extracted from impedance profiles and fluorescent were fused and then applied to support vector machines, producing a ~100% accuracy in cell-type classification.