Machine learning empowered multi-stress level electromechanical phenotyping for high-dimensional single cell analysis

Abstract

Microfluidics provides a powerful platform for biological analysis by harnessing the ability to precisely manipulate fluids and microparticles with integrated microsensors. Here, we introduce an imaging and impedance cell analyzer (IM2Cell), which implements single cell level impedance analysis and hydrodynamic mechanical phenotyping simultaneously. For the first time, IM2Cell demonstrates the capability of multi-stress level mechanical phenotyping. Specifically, IM2Cell is capable of characterizing cell diameter, three deformability responses, and four electrical properties. It presents high-dimensional information to give insight into subcellular components such as cell membrane, cytoplasm, cytoskeleton, and nucleus. In this work, we first validate imaging and impedance-based cell analyses separately. Then, the two techniques are combined to obtain both imaging and impedance data analyzed by machine learning method, exhibiting an improved prediction accuracy from 83.1% to 95.4% between fixed and living MDA-MB-231 breast cancer cells. Next, IM2Cell demonstrates 91.2% classification accuracy in a mixture of unlabeled MCF-10A, MCF-7, and MDA-MB-231 cell lines. Finally, an application demonstrates the potential of IM2Cell for the deformability studies of peripheral blood mononuclear cells (PBMCs) subpopulations without cumbersome isolation or labeling steps.