Back to search

University of Alabama Birmingham

Wavelet-Based Regression and Classification for Longitudinal Diffusion Tensor Imaging Data

Abstract

dc:description.abstract

Diffusion tensor imaging (DTI) is a magnetic resonance imaging (MRI) technique capable of in vivo characterization of the spatial and angular dependence of free water diffusion in tissue. By studying intra- and extra-cellular water mobility, inference can be made about the surrounding medium. The non-uniform angular dependence of diffusion, known as anisotropy, is evident in fibrous tissues, which exhibit greater diffusion parallel versus perpendicular to the fiber orientation. DTI’s sensitivity to anisotropy makes it an effective tool for measuring the integrity of fibrous myelinated white matter tracts in the brain. For demyelinating diseases such as multiple sclerosis (MS), DTI has seen extensive use for evaluating disease progression. Several techniques have been developed to analyze the significance of differences between several sets of images. There have been fewer attempts to automate the regression of clinical outcomes using DTI data. I consider the problem of quantifying the connection between observed differences in images and disease related clinical outcomes. I develop a functional linear model approach to automated generation of imaging biomarkers from DTI data. This method is implemented via a two stage model fitting procedure that begins with a wavelet-based voxelwise analysis to extract features related to the outcome. The second stage uses these features to predict the clinical outcome. The method is applied to the analysis of iv longitudinal DTI data from an MS cohort. A comprehensive software suite is developed in R to implement the method.

Degree

thesis:*
Level thesis:degree_level
Dissertation

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Prucka, William R
Contributors dc:contributor
  • Christopher S Coffey
  • Gary R Cutter
  • Charles R Katholi
  • N Shastry Akella
  • Donald B Twieg

Subjects

dc:subject × 1

Identifiers

dc:identifier.*

Chain of custody

source
Harvested from
University of Alabama Birmingham
Base URL
digitalcommons.library.uab.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Prucka, William R. Wavelet-Based Regression and Classification for Longitudinal Diffusion Tensor Imaging Data. Dissertation thesis, https://digitalcommons.library.uab.edu/etd-collection/280