Author Date

2026-6

Degree Name

Department

Physics and Astronomy

College

Computational, Mathematical and Physical Sciences

Defense Date

2026-04-27

Publication Date

2026-06-09

First Faculty Advisor

Gus Hart

First Faculty Reader

Bryan Morse

Honors Coordinator

Karine Chesnel

Keywords

cell locomotion, cryogenic electron tomography, machine learning, flagellar motors, computer vision

Abstract

With advances in cryogenic electron tomography, the ability to study bacterial structures in their cellular context has improved. However, 3D images of bacteria, called tomograms, have a low signal-to-noise ratio. This makes annotating structures of interest difficult, as traditional computer vision models struggle with tomograms and manual annotation is time consuming. We build on the results of the BYU Kaggle competition to create an ensemble model capable of automatically annotating flagellar models with nearly around 85% accuracy.