Abstract
Microfluidic impedance flow cytometry has been widely used in leukocyte differential and counting, but it faces a bottleneck due to the trade-off between impedance detection throughput and sensitivity. In this study, a microfluidic impedance flow cytometer based on a virtual constriction microchannel was reported, in which the virtual constriction microchannel was constructed by crossflow of conductive sample and insulated sheath fluids with underneath micro-electrodes for impedance measurements. Compared to conventional mechanical constriction microchannels, this virtual counterpart could effectively avoid direct physical contact between cells and the microchannel walls to maintain high throughputs, and significantly reduce the volume of the impedance detection region for sensitivity improvements. Using the developed microfluidic impedance flow cytometer, impedance pulses of three leukemia cell lines, K562, Jurkat, and HL-60, were detected, achieving a 99.8% differentiation accuracy through the use of a recurrent neural network. Furthermore, impedance pulses of four white blood cell subpopulations (neutrophils, eosinophils, monocytes, and lymphocytes) from three donors were detected, achieving a classification accuracy of ≥99.2%. A classification network model was established based on purified white blood cell and applied to impedance pulses of two white blood cell mixtures, resulting in proportional distributions of four leukocyte subpopulations within theoretical ranges. These results indicated that the developed microfluidic impedance flow cytometer based on the virtual constriction microchannel could achieve both high detection throughput and high sensitivity, showing great potentials for clinical diagnostics and blood analysis.