Physicochemical Profiling of Macrophage Heterogeneity Using Deep Learning Integrated Nanosensor Cytometry

Label-free single-cell analytics have been developed for understanding the collective immune response mechanism of immune cells. However, it remains difficult to analyze the physicochemical properties of a single cell in high spatiotemporal resolution for an immune cell having dynamic morphological changes and significant molecular heterogeneities. It is deemed due to the absence of a sensitive molecular sensing construct and single-cell imaging analytic program. In this study, we developed a deep learning integrated nanosensor chemical cytometry (DI-NCC) platform, which combines a fluorescent nanosensor array in microfluidics and a deep learning model for cell feature analysis. The DI-NCC platform possesses the capability to collect rich, multivariate data sets for each individual immune cell (e.g., macrophage) within the population. We obtained LPS+ (n = 25) and LPS- (n = 61) near-infrared images and analyzed 250 cells/mm² in 1 μm spatial resolution and 0 to 1.0 confidence level even with overlapped or adhered cell configurations. This enables automatic quantification of the activation and nonactivation levels of a single macrophage upon instantaneous immune stimulations. Furthermore, we support the activation level quantified by deep learning with heterogeneities analysis of both biophysical (cell size) and biochemical (nitric oxide efflux) properties. The DI-NCC platform can be promising for activation profiling of dynamic heterogeneity variations of cell populations.