Author

Fan Yang

ORCID Identifier(s)

0000-0002-4974-9560

Graduation Semester and Year

2022

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Computer Science

Department

Computer Science and Engineering

First Advisor

Hwa Won Kim

Second Advisor

Gautam Das

Third Advisor

Dajiang Zhu

Fourth Advisor

Amal Isaiah

Abstract

With the recent advance and widespread adoption of imaging technological innovations, clinical practitioner and scientists can easily acquire and store a large amount of various neuroimaging modalities, such as Diffusion Tensor Imaging (DTI), Magnetic Resonance Imaging (MRI), resting-state functional MRI (rs-fMRI) and Positron Emission Tomography (PET), etc. These novel imaging data sources cover a rich amount of factors that influence patients' cognitive health, offer an objective view of patients at unprecedented multi-resolution for the understanding of brain structure and function, and have the significant potential to improve healthcare by aiding better decision-making in diagnosing, monitoring and treating diseases. Machine Learning methods have emerged as the state-of-the-art in learning from the large-scale neuroimaging data. While their use for medical applications is interesting and insightful, it is often very challenging in practice. Some of the major challenges we encounter in the adoption of Machine Learning methods for neuroscience tasks are that examining the association between the socioeconomic characteristic and brain clinical measurements is difficult given the subtle variations between groups with different socioeconomic status, that effectively characterizing the early symptoms of the Alzheimer’s disease (AD) is in many cases not possible, that forecasting and capturing the disease-related dynamics of clinical measurements is necessary to better understand the progression of AD, and that modelling the dynamic associations between lengthy sequences of multivariate variables for brain connectivity analysis is computational expensive. To take care of these challenges, we propose multiple novel Machine Learning methods for providing a multi-scale representation of the original measurement to enhance the sensitivity of downstream statistical analysis, for integrating graph structure and diagnostic label information to characterize early symptoms of AD, for incorporating time-dependent label information to better understand the progression of AD, and for efficiently estimate and predict dynamic covariances on large-scale time series data. We demonstrate our developed methods on the challenging real-word data from various clinical studies in the neuroscience domain, including Adolescent Brain Cognitive Development (ABCD) study, Alzheimer’s Disease Neuroimaging Initiative (ADNI) study, and Human Connectome Project (HCP). Our contributions advance the state-of-the art in regard to leveraging Machine Learning methods for neuroscience applications and accentuate the foreground in which artificial intelligence on large-scale neuroimaging data can improve healthcare with better decision-makings.

Keywords

Machine learning, Neuroimaging, Statistical analysis, Representation learning

Disciplines

Computer Sciences | Physical Sciences and Mathematics

Comments

Degree granted by The University of Texas at Arlington

30396-2.zip (58943 kB)

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.