Physical Sciences and Mathematics Commons^™

Evaluating Neuroimaging Modalities In The A/T/N Framework: Single And Combined Fdg-Pet And T1-Weighted Mri For Alzheimer’S Diagnosis, Peiwang Liu

McKelvey School of Engineering Theses & Dissertations

With the escalating prevalence of dementia, particularly Alzheimer's Disease (AD), the need for early and precise diagnostic techniques is rising. This study delves into the comparative efficacy of Fluorodeoxyglucose Positron Emission Tomography (FDG-PET) and T1-weighted Magnetic Resonance Imaging (MRI) in diagnosing AD, where the integration of multimodal models is becoming a trend. Leveraging data from the Alzheimer's Disease Neuroimaging Initiative (ADNI), we employed linear Support Vector Machines (SVM) to assess the diagnostic potential of these modalities, both individually and in combination, within the AD continuum. Our analysis, under the A/T/N framework's 'N' category, reveals that FDG-PET consistently outperforms T1w-MRI across …

Investigating Applications Of Deep Learning For Diagnosis Of Post Traumatic Elbow Disease, Hugh James

McKelvey School of Engineering Theses & Dissertations

Traumatic events such as dislocation, breaks, and arthritis of musculoskeletal joints can cause the development of post-traumatic joint contracture (PTJC). Clinically, noninvasive techniques such as Magnetic Resonance Imaging (MRI) scans are used to analyze the disease. Such procedures require a patient to sit sedentary for long periods of time and can be expensive as well. Additionally, years of practice and experience are required for clinicians to accurately recognize the diseased anterior capsule region and make an accurate diagnosis. Manual tracing of the anterior capsule is done to help with diagnosis but is subjective and timely. As a result, there is …

Development Of The Assessment Of Clinical Prediction Model Transportability (Apt) Checklist, Sean Chonghwan Yu

McKelvey School of Engineering Theses & Dissertations

Clinical Prediction Models (CPM) have long been used for Clinical Decision Support (CDS) initially based on simple clinical scoring systems, and increasingly based on complex machine learning models relying on large-scale Electronic Health Record (EHR) data. External implementation – or the application of CPMs on sites where it was not originally developed – is valuable as it reduces the need for redundant de novo CPM development, enables CPM usage by low resource organizations, facilitates external validation studies, and encourages collaborative development of CPMs. Further, adoption of externally developed CPMs has been facilitated by ongoing interoperability efforts in standards, policy, and …

Integrating Physical Models And Deep Priors For Computational Imaging, Yu Sun

McKelvey School of Engineering Theses & Dissertations

This dissertation addresses integrating physical models and learning priors for computational imaging. The motivation of our work is driven by the recent discussion of learning-based methods that solve the imaging inverse problem by directly learning a measurement-to-image mapping from the existing data: they achieve superior performance over the traditional model-based methods but lack the physical model to impose sufficient interpretation and guarantee of the final image. We adopt the classic statistical inference as the underlying formulation and integrate learning models as implicit image priors, such that our framework is able to simultaneously leverage physical models and learning priors. Additionally, the …

Machine Learning In Complex Scientific Domains: Hospitalization Records, Drug Interactions, Predictive Modeling And Fairness For Class Imbalanced Data, Arghya Datta

McKelvey School of Engineering Theses & Dissertations

Machine learning has demonstrated potential in analyzing large, complex datasets and has become ubiquitous across many fields of scientific research. As machine learning is actively deployed in many complex and critical domains, it is essential for machine learning to engage with domain expertise to aid in knowledge discovery as well as address challenges in predictive modeling in complex domains. Domain expertise represents an essential and elaborate collection of knowledge that is often under-utilized when applying machine learning in complex domains. In this dissertation, I have addressed existing challenges regarding knowledge discovery in complex domains via engagement with domain expertise, particularly …

Continuous-Time And Complex Growth Transforms For Analog Computing And Optimization, Oindrila Chatterjee

McKelvey School of Engineering Theses & Dissertations

Analog computing is a promising and practical candidate for solving complex computational problems involving algebraic and differential equations. At the fundamental level, an analog computing framework can be viewed as a dynamical system that evolves following fundamental physical principles, like energy minimization, to solve a computing task. Additionally, conservation laws, such as conservation of charge, energy, or mass, provide a natural way to couple and constrain spatially separated variables. Taking a cue from these observations, in this dissertation, I have explored a novel dynamical system-based computing framework that exploits naturally occurring analog conservation constraints to solve a variety of optimization …

Reasoning About Scene And Image Structure For Computer Vision, Zhihao Xia

McKelvey School of Engineering Theses & Dissertations

The wide availability of cheap consumer cameras has democratized photography for novices and experts alike, with more than a trillion photographs taken each year. While many of these cameras---especially those on mobile phones---have inexpensive optics and make imperfect measurements, the use of modern computational techniques can allow the recovery of high-quality photographs as well as of scene attributes.

In this dissertation, we explore algorithms to infer a wide variety of physical and visual properties of the world, including color, geometry, reflectance etc., from images taken by casual photographers in unconstrained settings. We specifically focus on neural network-based methods, while incorporating …

Machine Learning Morphisms: A Framework For Designing And Analyzing Machine Learning Work Ows, Applied To Separability, Error Bounds, And 30-Day Hospital Readmissions, Eric Zenon Cawi

McKelvey School of Engineering Theses & Dissertations

A machine learning workflow is the sequence of tasks necessary to implement a machine learning application, including data collection, preprocessing, feature engineering, exploratory analysis, and model training/selection. In this dissertation we propose the Machine Learning Morphism (MLM) as a mathematical framework to describe the tasks in a workflow. The MLM is a tuple consisting of: Input Space, Output Space, Learning Morphism, Parameter Prior, Empirical Risk Function. This contains the information necessary to learn the parameters of the learning morphism, which represents a workflow task. In chapter 1, we give a short review of typical tasks present in a workflow, as …

Mapping Transcription Factor Networks And Elucidating Their Biological Determinants, Yiming Kang

McKelvey School of Engineering Theses & Dissertations

A central goal in systems biology is to accurately map the transcription factor (TF) network of a cell. Such a network map is a key component for many downstream applications, from developmental biology to transcriptome engineering, and from disease modeling to drug discovery. Building a reliable network map requires a wide range of data sources including TF binding locations and gene expression data after direct TF perturbations. However, we are facing two roadblocks. First, rich resources are available only for a few well-studied systems and cannot be easily replicated for new organisms or cell types. Second, when TF binding and …

Differential Estimation Of Audiograms Using Gaussian Process Active Model Selection, Trevor Larsen

McKelvey School of Engineering Theses & Dissertations

Classical methods for psychometric function estimation either require excessive resources to perform, as in the method of constants, or produce only a low resolution approximation of the target psychometric function, as in adaptive staircase or up-down procedures. This thesis makes two primary contributions to the estimation of the audiogram, a clinically relevant psychometric function estimated by querying a patient’s for audibility of a collection of tones. First, it covers the implementation of a Gaussian process model for learning an audiogram using another audiogram as a prior belief to speed up the learning procedure. Second, it implements a use case of …

Knowledge Driven Approaches And Machine Learning Improve The Identification Of Clinically Relevant Somatic Mutations In Cancer Genomics, Benjamin John Ainscough

Arts & Sciences Electronic Theses and Dissertations

For cancer genomics to fully expand its utility from research discovery to clinical adoption, somatic variant detection pipelines must be optimized and standardized to ensure identification of clinically relevant mutations and to reduce laborious and error-prone post-processing steps. To address the need for improved catalogues of clinically and biologically important somatic mutations, we developed DoCM, a Database of Curated Mutations in Cancer (http://docm.info), as described in Chapter 2. DoCM is an open source, openly licensed resource to enable the cancer research community to aggregate, store and track biologically and clinically important cancer variants. DoCM is currently comprised of 1,364 variants …

On The Aggregation Of Subjective Inputs From Multiple Sources, Mithun Chakraborty

McKelvey School of Engineering Theses & Dissertations

When we have a population of individuals or artificially intelligent agents possessing diverse subjective inputs (e.g. predictions, opinions, etc.) about a common topic, how should we collect and combine them into a single judgment or estimate? This has long been a fundamental question across disciplines that concern themselves with forecasting and decision-making, and has attracted the attention of computer scientists particularly on account of the proliferation of online platforms for electronic commerce and the harnessing of collective intelligence. In this dissertation, I study this problem through the lens of computational social science in three main parts: (1) Incentives in information …

Visualization Of Deep Convolutional Neural Networks, Dingwen Li

McKelvey School of Engineering Theses & Dissertations

Deep learning has achieved great accuracy in large scale image classification and scene recognition tasks, especially after the Convolutional Neural Network (CNN) model was introduced. Although a CNN often demonstrates very good classification results, it is usually unclear how or why a classification result is achieved. The objective of this thesis is to explore several existing visualization approaches which offer intuitive visual results. The thesis focuses on three visualization approaches: (1) image masking which highlights the region of image with high influence on the classification, (2) Taylor decomposition back-propagation which generates a per pixel heat map that describes each pixel's …

Learning With Scalability And Compactness, Wenlin Chen

McKelvey School of Engineering Theses & Dissertations

Artificial Intelligence has been thriving for decades since its birth. Traditional AI features heuristic search and planning, providing good strategy for tasks that are inherently search-based problems, such as games and GPS searching. In the meantime, machine learning, arguably the hottest subfield of AI, embraces data-driven methodology with great success in a wide range of applications such as computer vision and speech recognition. As a new trend, the applications of both learning and search have shifted toward mobile and embedded devices which entails not only scalability but also compactness of the models. Under this general paradigm, we propose a series …

Application Of Machine Learning To Mapping And Simulating Gene Regulatory Networks, Hien-Haw Liow

Arts & Sciences Electronic Theses and Dissertations

This dissertation explores, proposes, and examines methods of applying modernmachine learning and Bayesian statistics in the quantitative and qualitative modeling of gene regulatory networks using high-throughput gene expression data. A semi-parametric Bayesian model based on random forest is developed to infer quantitative aspects of gene regulation relations; a parametric model is developed to predict geneexpression levels solely from genotype information. Simulation of network behavior is shown to complement regression analysis greatly in capturing the dynamics of gene regulatory networks. Finally, as an application and extension of novel approaches in gene expression analysis, new methods of discovering topological structure of gene …