Defense Date

3-11-2024

Graduation Date

Spring 5-10-2024

Availability

One-year Embargo

Submission Type

thesis

Degree Name

MS

Department

Biomedical Engineering

School

School of Science and Engineering

Committee Chair

Melikhan Tanyeri

Committee Member

John Viator

Committee Member

Bin Yang

Keywords

Microfluidics, Bacteria, Machine Learning, SVM, Bacteria Enumeration

Abstract

The capacity to identify bacteria quickly and accurately is critical for applications such as medical diagnosis, environmental monitoring, and food safety. For instance, sepsis diagnosis requires blood cultures that take as long as 1-5 days, preventing timely intervention and increasing mortality rate. Here, we propose the joint use of microfluidics and a machine learning algorithm for rapid bacterial cell capture and quantification. The combination of high-throughput droplet microfluidics and a support vector machine (SVM) enables analysis and quantification of bacterial samples within as short as 4 hours. We have performed successful encapsulation of pathogenic bacteria such as E. coli, S. enterica, and P. aeruginosa, yielding an average detection accuracy of 98% using our trained SVM model. These results were benchmarked with traditional plating techniques and Petroff-Hausser chamber counts, demonstrating comparable growth curves and validating the promise of digital bacterial enumeration methods.

Language

English

Download

Share

COinS