Show simple item record

dc.contributor.authorZhao, Yu
dc.date.accessioned2019-02-19T05:30:16Z
dc.date.available2019-02-19T05:30:16Z
dc.date.issued2018-12
dc.identifier.otherzhao_yu_201812_phd
dc.identifier.urihttp://purl.galileo.usg.edu/uga_etd/zhao_yu_201812_phd
dc.identifier.urihttp://hdl.handle.net/10724/38595
dc.description.abstractMedical image analysis plays an important role for understanding both human psychology and physiology. Human body functional and structural information can be recorded using different medical imaging techniques. As for functional image analysis, understanding the organizational architecture of human brain function has been of intense interest since the inception of human neuroscience. In-vivo Functional Magnetic Resonance Imaging (fMRI) technology enabled the investigation of the human brain functional mechanism and the decomposition of the brain functional network components. However, it is a great challenge to conduct simultaneous spatial-temporal decomposition analysis due to the 4D nature of the fMRI data. Existing methods, i.e., general linear model, independent component analysis, sparse representation, have been proposed for purely functional network decomposition on either spatial or temporal feature of fMRI data. Spatial functional network analysis after fMRI decomposition is still an open question. The major challenge is the lack of ability to effectively describe spatial volume maps of brain networks exposed to extensive individual variability. Besides, the 4D nature of fMRI data has not been fully investigated due to the traditional methodology limitation. To address the abovementioned challenges, we proposed a series of deep learning frameworks for spatial functional network map descriptor and 4D spatial temporal analysis of fMRI data. Applications include both task-evoked fMRI, resting state fMRI and also brain disease fMRI data. As for structural image analysis, magnetic resonance (MR) images and computed tomography (CT) are both primary structural imaging modalities for solving and analyzing various medical problems. Understanding of human body structure is advanced by medical imaging technologies, while medical images analysis is advanced by the development of new digital analysis techniques from computer science community. To provide better solutions for some existing challenging medical problems, we proposed deep learning frameworks to facilitate cross modality synthesis and landmark detection tasks, both are widely used and eagerly needed in healthcare industry.
dc.languageeng
dc.publisheruga
dc.rightspublic
dc.subjectfunctional magnetic resonance imaging analysis, structural medical imaging analysis
dc.subjectspatial temporal analysis
dc.subjectcross modality analysis
dc.subjectdeep learning
dc.titleDeep learning frameworks for functional and structural medical image analysis
dc.typeDissertation
dc.description.degreePhD
dc.description.departmentComputer Science
dc.description.majorComputer Science
dc.description.advisorTianming Liu
dc.description.committeeTianming Liu
dc.description.committeeShannon Quinn
dc.description.committeeSuchendra Bhandarkar


Files in this item

FilesSizeFormatView

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record