Date of Award

8-2026

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical and Computer Engineering

Committee Chair/Advisor

Pingshan Wang

Committee Member

Goutam Koley

Committee Member

Rod Harrell

Committee Member

Matt Turnbull

Abstract

Chinese hamster ovary (CHO) cells and other cultured cell lines are the workhorses of biopharmaceutical manufacturing, yet the tools available to monitor the state of individual cells during culture remain limited: most are invasive, destructive, low in throughput, or dependent on fluorescent or chemical labels. A single measurement able to report cell growth, physiological stress, metabolic activity, and cell-line identity---label-free, non-invasively, and one cell at a time---would be broadly useful for cell-line development and bioprocess monitoring. This dissertation develops such a sensor system: a microstrip-line microwave dielectric spectroscopic cytometer (DiSC), also termed as a microwave flow cytometer (MFC), that records the complex scattering parameters (S11 and S21) of single cells suspended in their culture medium. At gigahertz frequencies the probing field penetrates the cell membrane and interacts with intracellular ions and molecules, so the measured signal reflects the cell's molecular composition profile and its biomass. Because features are extracted as event-relative changes (ΔS) referenced to each cell's own local baseline, the measurement is reproducible, insensitive to cell position and orientation, and robust to slow drift---without labels or immobilization. Three studies demonstrate the platform across complementary axes of cell state. First, continuous single-cell measurement of budding yeast (Saccharomyces pastorianus) captures biomass accumulation and the heat-shock response in real time, resolving segmented growth, rapid mass fluctuations, heterogeneous onset of apoptosis, and divergent outcomes among closely related progeny. Second, population- and single-cell measurements of CHO cells exposed to the mitochondrial inhibitors oligomycin and FCCP show that the sensor detects cellular responses to metabolic perturbation---including dynamic single-cell responses tracked continuously for approximately sixty minutes---while inert polystyrene beads bracketing each measurement separate genuine cell responses from medium-level artifacts. Third, applied to cell-line discrimination, machine-learning classifiers trained on the microwave signatures separate most of seven CHO lines at high recall and, in a blind co-measured mixture, independently and consistently assign individual cells of an antibody-producing recombinant line versus its host line, demonstrating label-free single-cell discrimination of producer from host. Together these results establish microwave dielectric sensing, combined with machine learning, as a promising label-free, non-invasive, high-throughput approach for monitoring cell growth, stress, metabolic response, and cell-line identity in suspension culture. The work also delineates the current limits of the approach: the measurements reliably report relative dielectric perturbations rather than absolute intrinsic permittivity, whose accurate extraction requires rigorous de-embedding that remains an open problem. Extending the analysis to the full set of acquired frequencies, adding complementary features such as video-based cell sizing, validating single-cell assignments against orthogonal references, and integrating the sensor into at-line or in-bioreactor monitoring are identified as the natural next steps toward deployment in cell-line development and biomanufacturing.

Author ORCID Identifier

https://orcid.org/0009-0001-4044-8198

Download

Available for download on Tuesday, August 31, 2027

Share

COinS