Independent Component Analysis for Group Comparison of Functional MRI Images in Individuals with Parkinson's Disease
AdvisorBedrick, Edward J.
MetadataShow full item record
PublisherThe University of Arizona.
RightsCopyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction, presentation (such as public display or performance) of protected items is prohibited except with permission of the author.
AbstractParkinson's Disease (PD) is an age-related disorder that affects cognitive and motor abilities and lowers quality of life. As there is currently no cure, it is an area of interest for many research efforts. Parkinson's disease has a substantial effect on structures in the basal ganglia, which may be used to indicate signs of Parkinson's disease progression. Functional MRI (fMRI) is a means of measuring metabolic functioning in the brain. Brain imaging studies are not used to diagnose Parkinson's disease because it is unclear how it manifests in neuroimages. However, Parkinson's disease has a preclinical phase during which structures within the brain are affected, but external symptoms have not yet manifested. In this study, we sought to identify effects of Parkinson's disease that may be seen in functional imaging scans to allow earlier detection. We used independent component analysis (ICA) to identify functional brain networks followed by dual regression to estimate subject-level components for making comparisons of the functional images between the two groups: a group of subjects with Parkinson's disease and a group of healthy control subjects. We were able to generate subject-level components; however, identifying one component at the group level that included the basal ganglia proved problematic. Methods of identifying neural structures within the application we used provided conflicting evidence. Therefore, we were unable to determine if differences between the study groups existed that could be seen in functional imaging scans. Testing our primary endpoint using a voxel-wise linear regression with each of the components was not successful because most of the p values on the coefficient of interest were non-significant. In addition, there was a poor model fit seen in the regression models. We were unable to provide scientific evidence of differences that might be seen in functional MRI studies between subjects with Parkinson's disease and healthy control subjects.
Degree ProgramGraduate College