The Library
Functional data analysis approaches for 3-dimensional brain images
Tools
Palma, Marco (2021) Functional data analysis approaches for 3-dimensional brain images. PhD thesis, University of Warwick.
|
PDF
WRAP_Theses_Palma_2021.pdf - Submitted Version - Requires a PDF viewer. Download (28Mb) | Preview |
Official URL: http://webcat.warwick.ac.uk/record=b3765998
Abstract
Functional data analysis approaches for 3-dimensional brain images
The analysis of brain images poses many challenges from a statistical perspective. First, these images are usually high-dimensional (sometimes millions of data points for each image), therefore a statistical analysis based on scalable techniques is often required. Second, these data exhibit clear spatial dependence due to the differences in structures and functions of the brain regions. Functional data analysis is a modern branch of statistics aimed at analysing data that are in the form of functions. Many tools from multivariate analysis and nonparametric smoothing are used in functional data analysis to reduce noise and perform dimension reduction. This thesis shows three applications of functional data analysis for large-scale 3-dimensional brain images, mainly focusing on prediction of scalar and imaging outcomes. A workflow for building prediction intervals for scalar outcomes from 3D covariates is devised and applied for the prediction of individual chronological age from brain anatomical images. Then, a framework for the analysis of functional data with spatially-dependent mean-variance relationship and skewness is described, with an application to structural imaging. At last, a functional imaging problem is studied: the prediction of a task-evoked response image from resting-state data is achieved through an image-on-image regression model. The results discussed in this thesis are mostly comparable with more complicated machine-learning approaches available in the literature, while being more easily interpretable and often more computationally appealing. Functional data analysis might represent a valid option for the statistical analysis of brain images even in high-dimensional setting.
Item Type: | Thesis (PhD) | ||||
---|---|---|---|---|---|
Subjects: | Q Science > QA Mathematics Q Science > QH Natural history Q Science > QP Physiology R Medicine > RC Internal medicine |
||||
Library of Congress Subject Headings (LCSH): | Brain -- Imaging -- Statistical methods, Brain -- Aging -- Imaging -- Statistical methods, Functional analysis, Three-dimensional imaging in biology, Three-dimensional imaging in medicine | ||||
Official Date: | November 2021 | ||||
Dates: |
|
||||
Institution: | University of Warwick | ||||
Theses Department: | Department of Statistics | ||||
Thesis Type: | PhD | ||||
Publication Status: | Unpublished | ||||
Supervisor(s)/Advisor: | Tavakoli, Shahin ; Nichols, Thomas E. ; Brettschneider, Julia | ||||
Sponsors: | Engineering and Physical Sciences Research Council | ||||
Format of File: | |||||
Extent: | xiv, 120 leaves : illustrations, charts | ||||
Language: | eng |
Request changes or add full text files to a record
Repository staff actions (login required)
View Item |
Downloads
Downloads per month over past year