Oracle Compound Decision Rules for False Discovery Rate Control in fMRI studies

Simple item page
dc.contributor.advisorWegman, Edward J.
dc.contributor.authorChen, Nan
dc.creatorChen, Nan
dc.date2013-05
dc.date.accessioned2013-08-15T20:21:37Z
dc.date.available2018-06-01T06:41:32Z
dc.date.issued2013-08-15
dc.description.abstractThe recent advance on functional magnetic resonance imaging (fMRI) allows scientists to assess functionality of the brain by measuring the response of blood flow to one or multiple types of stimuli. The analysis of fMRI studies is a very challenging high dimensional problem due to the fact that the statistical analysis will be applied to the measurements from huge volume of voxels. Based on the most recent development of the oracle false discovery rate, this thesis proposes three new approaches to the analysis of the fMRI studies. They are in the areas of 1) noise suppression by wavelet based Oracle False Discovery Rate (OFDR) threshholding; 2) voxel-based signal extraction with Enhanced Oracle False Discovery Rate (EOFDR); 3) group-based signal extraction using Brain-Map-Constrained adaptive oracle False Discovery Rate Approach. The simulation studies show improved performance of the proposed approaches compared to some existing competitive approaches. The proposed approaches are applied to fMRI studies to demonstrate practical usage.
dc.description.noteThis work is embargoed by the author and will not be available until June 2018.
dc.identifier.urihttps://hdl.handle.net/1920/8308
dc.language.isoen_US
dc.rightsCopyright 2013 Nan Chen
dc.subjectDenoising
dc.subjectFalse discovery rate
dc.subjectFunctional Magnetic Resonance Imaging
dc.subjectGroup analysis
dc.subjectMultiple testing
dc.subjectWavelet
dc.titleOracle Compound Decision Rules for False Discovery Rate Control in fMRI studies
dc.typeDissertation
thesis.degree.disciplineComputational Sciences and Informatics
thesis.degree.grantorGeorge Mason University
thesis.degree.levelDoctoral
thesis.degree.namePhD in Computational Sciences and Informatics

Files

License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.65 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

College of Science