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


Wu, Minjie (2010) REGISTRATION AND SEGMENTATION OF BRAIN MR IMAGES FROM ELDERLY INDIVIDUALS. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Primary Text

Download (22MB) | Preview


Quantitative analysis of the MRI structural and functional images is a fundamental component in the assessment of brain anatomical abnormalities, in mapping functional activation onto human anatomy, in longitudinal evaluation of disease progression, and in computer-assisted neurosurgery or surgical planning. Image registration and segmentation is central in analyzing structural and functional MR brain images. However, due to increased variability in brain morphology and age-related atrophy, traditional methods for image registration and segmentation are not suitable for analyzing MR brain images from elderly individuals. The overall goal of this dissertation is to develop algorithms to improve the registration and segmentation accuracy in the geriatric population. The specific aims of this work includes 1) to implement a fully deformable registration pipeline to allow a higher degree of spatial deformation and produce more accurate deformation field, 2) to propose and validate an optimum template selection method for atlas-based segmentation, 3) to propose and validate a multi-template strategy for image normalization, which characterizes brain structural variations in the elderly, 4) to develop an automated segmentation and localization method to access white matter integrity (WMH) in the elderly population, and finally 5) to study the default-mode network (DMN) connectivity and white matter hyperintensity in late-life depression (LLD) with the developed registration and segmentation methods. Through a series of experiments, we have shown that the deformable registration pipeline and the template selection strategies lead to improved accuracy in the brain MR image registration and segmentation, and the automated WMH segmentation and localization method provides more specific and more accurate information about volume and spatial distribution of WMH than traditional visual grading methods. Using the developed methods, our clinical study provides evidence for altered DMN connectivity in LLD. The correlation between WMH volume and DMN connectivity emphasizes the role of vascular changes in LLD's etiopathogenesis.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairAizenstein, Howard Jaizen@pitt.eduAIZEN
Committee MemberAndreescu, Carmenandrcx@upmc.eduCAA8
Committee MemberStetten,
Committee MemberHuppert, Theodorehuppertt@upmc.eduHUPPERT1
Date: 26 January 2010
Date Type: Completion
Defense Date: 17 November 2009
Approval Date: 26 January 2010
Submission Date: 30 November 2009
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Institution: University of Pittsburgh
Schools and Programs: Swanson School of Engineering > Bioengineering
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: image registration; image segmentation; late-life depression.; resting state connectivity; template selection
Other ID:, etd-11302009-184916
Date Deposited: 10 Nov 2011 20:07
Last Modified: 19 Dec 2016 14:37


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item