Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Spatio-Temporal Mixed Models for Diffusion Tensor Magnetic Resonance Imaging

Scott, John A. (2008) Spatio-Temporal Mixed Models for Diffusion Tensor Magnetic Resonance Imaging. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

[img]
Preview
PDF
Primary Text

Download (2MB) | Preview

Abstract

Diffusion tensor imaging (DTI) is a magnetic resonance imaging modality that provides useful in vivo information about the microstructure of human brain tissue, particularly the white matter structures that comprise the 'wiring' of the brain. DTI holds great promise for enhancing our understanding of white matter disorders, which comprise public health burdens in a variety of medical domains. Due to its relatively complex structure, however, extracting useful information from DTI data presents a number of statistical challenges. More effective statistical methodologies will improve the sensitivity of DTI data analyses and increases their clinical relevance, a goal of substantial public health significance. In this dissertation, I propose a series of analytic approaches to DTI data analysis based on linear mixed effects models (LMEs). These models provide a number of advantages over several expedient DTI data analyses in current use. I demonstrate the applicability and advantages of my LME-based approaches in an analysis that compares white matter microstructure in a group of children and young adults with autism spectrum disorders (ASDs) to typically developing controls.I first identify a class of LMEs for DTI data analyses for which closed-form maximum likelihood estimators of all parameters exist. By avoiding iteration, these models enable practitioners to perform exploratory and confirmatory analyses of large DTI datasets in clinically feasible time. This family of models incorporates group heterogeneity in variance-covariance structure. I then compare the results of my approach with other approaches currently in practice in an analysis of white matter abnormalities associated with ASDs. I also introduce a data analytic framework that incorporates the entire multivariate tensor in a single analysis. I last describe a unified likelihood-based approach to addressing reliability with a new estimator of a generalized intraclass correlation coefficient. I establish the robustness of this approach to model perturbations with a series of theoretical and simulation results and apply it to quantify local spatial reliability in the ASDs example.


Share

Citation/Export:
Social Networking:
Share |

Details

Item Type: University of Pittsburgh ETD
Status: Unpublished
Creators/Authors:
CreatorsEmailPitt UsernameORCID
Scott, John A.johnascott@gmail.com
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairLange, Nicholasnlange@hms.harvard.edu
Committee MemberSiegle, Greggsiegle@pitt.eduGSIEGLE
Committee MemberCho, Raymondchory@upmc.edu
Committee MemberDay, Rogerday@upci.pitt.eduDAY01
Committee MemberAnderson, Stewartsja@pitt.eduSJA
Date: 28 September 2008
Date Type: Completion
Defense Date: 8 August 2008
Approval Date: 28 September 2008
Submission Date: 1 August 2008
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Institution: University of Pittsburgh
Schools and Programs: Graduate School of Public Health > Biostatistics
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: biostatistics; diffusion tensor imaging; DT-MRI; DTI; lme; mixed models; reliability; statistics
Other ID: http://etd.library.pitt.edu/ETD/available/etd-08012008-165807/, etd-08012008-165807
Date Deposited: 10 Nov 2011 19:56
Last Modified: 15 Nov 2016 13:47
URI: http://d-scholarship.pitt.edu/id/eprint/8805

Metrics

Monthly Views for the past 3 years

Plum Analytics


Actions (login required)

View Item View Item