Masters Theses
Date of Award
8-2020
Degree Type
Thesis
Degree Name
Master of Science
Major
Biochemistry and Cellular and Molecular Biology
Major Professor
Hong Guo
Committee Members
Tian Hong, Michael A Langston, Haileab Hilafu, Tongye Shen
Abstract
Cells can display a diverse set of motility behaviors, and these behaviors may reflect a cell’s functional state. Automated, and accurate cell motility analysis is essential to cell studies where the analysis of motility pattern is required. The results of such analysis can be used for diagnostic or curative decisions. Deep learning area has made astonishing progresses in the past several years. For computer vision tasks, different convolutional neural networks (CNN) and optimizers have been proposed to fix some problems. For time sequence data, recurrent neural networks (RNN) have been widely used.
This project leveraged on these recent advances to find the proper neural network for bacterial motility trajectory analysis for genotypic classification. This thesis was trying to answer two questions: (1) Which machine learning model can effectively classify the genotype of bacterial based on their motility patterns? And (2) Which motility parameters can best predict the bacterial genotype? The first question is addressed in the result 1 and 2 using different data formats and different machine learning models. The second question is addressed in result 3. Accordingly, this thesis is divided into three parts: (1) different traditional machine learning models are tested for predicting bacterial genotype using the coordinates’ sequences extracted from microscopic videos. (2) bacterial genotype classification task is solved by using deep neural network and the raw videos. (3) different motility parameters are tested to find out the best for predicting bacterial genotypes.
It is found that neural network gives highest accuracy in classifying bacterial genotype using coordinates’ sequences. Deep neural network with CNN-RNN can effectively classify the bacterial genotype using video data. Among popular motility parameters, some of them predict the bacterial genotype 20% better than others. The broader impact of this project is to automate trajectory analysis process and enable high-throughput trajectory analysis for research and clinical uses.
Recommended Citation
Ma, Yue, "Classification of bacterial motility using machine learning. " Master's Thesis, University of Tennessee, 2020.
https://trace.tennessee.edu/utk_gradthes/6258