Statistical Methods for Functional Magnetic Resonance Imaging Data

fMRI;imaging;shrinkage;empirical Bayes;outlier detection;resting-state;functional connectivity;Biostatistics

Mejia, AmandaPekar, James J. ;

Johns Hopkins University

关键词: fMRI;    imaging;    shrinkage;    empirical Bayes;    outlier detection;    resting-state;    functional connectivity;    Biostatistics;   

Others  :  https://jscholarship.library.jhu.edu/bitstream/handle/1774.2/40725/root.tex?sequence=8&isAllowed=y

瑞士|英语

来源: JOHNS HOPKINS DSpace Repository

【 摘 要 】

Understanding how the brain functions is one of the most important goals in science and medicine today.Functional magnetic resonance imaging (fMRI) is a noninvasive, widely used technology for studying brain function in humans.While fMRI has great potential to shed light on cognitive development, decline and disorders, it also presents statistical and computational challenges due to a myriad of sources of noise and the large size of the data.In this thesis, I propose several methods to improve the analysis of resting-state fMRI, which is used to understand connectivity between different regions of the brain.Specifically, this thesis addresses two primary themes.First, I propose shrinkage estimators for functional connectivity, which improve reliability of subject-level estimates by ;;borrowing strength;; across subjects.Second, I propose a method of identifying artifacts in fMRI data through a novel high-dimensional outlier detection method.The proposed methods can be used together and have the potential to significantly improve our understanding of brain connectivity at the subject level using resting-state fMRI.