Smartphone-Based Detection of Natural Killer Cells Using Flow-Based Measurement and Machine Learning Classification on Paper Microfluidics
Publisher
The University of Arizona.Rights
Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction, presentation (such as public display or performance) of protected items is prohibited except with permission of the author.Abstract
Natural Killer (NK) cells are innate immune cells capable of cytotoxic activity that defends against viral infections and cancer, and as such, these cells have become attractive as cancer immunotherapies. Subpopulations of NK cells include CD56bright and CD56dim NK’s that perform either cytokine production or direct cytotoxic cell killing respectively, and the absolute number and proportion of these cells in peripheral blood is important in maintaining immune function. Current methods of performing analysis of the cytokine environment as well as the number and proportion of NK cell subpopulations includes the use of immunoassays, flow cytometry, and numerous fluorescent dyes, as well as highly specialized equipment. We have developed a smartphone based device for the prognostics of engineered NK cell therapy using a two component flow based paper microfluidic chip and machine learning classification. The first unit composed of grade 1 chromatography paper measures flow velocity via video to provide information on cytokine and total NK cell concentrations in undiluted buffy coat. The second, single flow lane unit performs spatial separation of CD56bright and CD56dim cells over its length using differential binding of anti-CD56 nanoparticles. A smartphone based fluorescent microscope was developed in combination with an graphical user interface to perform analysis of flow data as well as analysis of subpopulations via machine learning. Limits of detection for cytokine and cell concentrations were found to be 50IU and 1000 cells/mL respectively while classification accuracy for cell subpopulations was found to be 88%. Our device is capable of enabling prognostics for NK cell therapy from complex buffy coat samples in a uniquely low cost, easy to use, and point-of-care fashion.Type
textElectronic Thesis
Degree Name
M.S.Degree Level
mastersDegree Program
Graduate CollegeBiomedical Engineering