Date of Award

8-2024

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Bioengineering

Committee Chair/Advisor

Dr. Bruce Z. Gao

Committee Member

Dr. David Karig

Committee Member

Dr. Lucas Schmidt

Committee Member

Dr. Tong Ye

Abstract

Hyperspectral imaging is a non-invasive imaging method capable of collecting both spatial and spectral information. However, because of the large volume of data collected, much of it is redundant or not useful for classification. Deep learning is a subset of machine learning that uses artificial neurons in a multilayered structure to learn representations from data. One of the main advantages of deep learning is the powerful feature extraction capabilities, which allow the model to learn both high- and low-level features. Convolutional neural networks are a type of deep learning model that have alternating convolutional and pooling layers capable of extracting features while also downsampling the input image to remove unimportant features, retain important features and reduce the computational load of the deep learning model.

CNNs can be combined with hyperspectral imaging to rapidly process the large amounts of data while removing unnecessary features. Thus, CNNs can be used to classify different hyperspectral images of bacteria and determine which bacteria are present in the image based on their spectral responses. From the output of the network, false color images can be generated from the input hyperspectral images to enhance visualization and provide spatial information.

In this thesis, we explained the goal of this project to design a deep learning model capable of processing hyperspectral images of bacteria and outputting classifications with high accuracy. We described a detailed procedure of the development of a three-layer convolutional neural network capable of generating predictions with approximately 96% accuracy and producing false color images of the hyperspectral inputs.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.