Skip to Main Content
 

Global Search Box

 
 
 
 

Files

ETD Abstract Container

Abstract Header

A semiparametric statistical approach to Functional MRI data

Abstract Details

2009, Doctor of Philosophy, Ohio State University, Statistics.

Functional magnetic resonance imaging measures the Hemodynamic Response (HR), to help understand the relationship between activity in certain brain areas and specific cognitive functions. Although recent breakthroughs in imaging techniques have led to an improved understanding of brain-behavior relationships, cognitive neuroscientists have struggled with developing an understanding of the complicated neural connections across regions in the brain possibly accompanied by different delays in responses to different tasks, as well as a high variation in brain activation between subjects.

In this study, a semiparametric statistical analysis procedure, utilizing a parametric assumption on task-related HR patterns and nonparametric statistical analysis, is proposed to resolve the issues of high variability in neural behavior across regions, subjects, and tasks. The proposed procedure consists of two parts: within-subject analysis and between-subject analysis. In within-subject analysis, temporal domain analysis, which refers to an estimation method for a task-related convolved HRF pattern from each subject and each experimental condition, namely temporal domain analysis,is proposed to consider high variability in HRF between subjects, regions, and tasks. Within-subject analysis includes spatial domain analysis to estimate individual spatial activation map. Within-subject analysis is repeated for each subject and each experimental condition and the results, obtained from within-subject analysis, are then tested for voxel-wise inference in between-subject analysis.

For an application of the proposed method, data from a fMRI experiment consisting of two cognitive tasks, phonological (relationship by sound) and semantic (relationship by meaning), and two drug conditions, L-dopa and placebo, with 16 subjects, were used. For each task, subjects were presented with a cue word and a list of words and directed to respond by pressing YES or NO buttons, to indicate whether or not a word in the list was related to the cue word.

Activated regions in the brain, found in this study, for the two cognitive tasks with placebo condition include: the left inferior frontal cortex; the fusiform gyrus; the posterior superior and middle temporal gyrus; the left parietal lobe. These areas are known to be highly task-related regions. In both tasks, administrating L-dopa resulted in more activation in a posterior region, superior and middle temporal gyrus compared to administrating placebo. The pulvinar thalamus showed higher activation with placebo than with L-dopa in the phonological task. Such posterior effects with L-dopa in the phonological task may be explained with less activation in the pulvinar with L-dopa. Further more, such L-dopa effects on posterior region in the brain may be due to the effects of L-dopa on semantic priming test (Kischka et al., 1996).

The effectiveness of the proposed procedure is assessed by a comparison between results from the proposed method and results from traditional General Linear Model (GLM) analysis. Both methods found similar group activation maps for each task and drug condition. However, in comparison between L-dopa and placebo for each task condition, the GLM method found no significant differences.

Prem Goel, K. (Advisor)
David Beversdorf, Q. (Committee Member)
Douglas Wolfe, A. (Committee Member)
Thomas Santner, J. (Committee Member)
122 p.

Recommended Citations

Citations

  • KIM, N. (2009). A semiparametric statistical approach to Functional MRI data [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1262295445

    APA Style (7th edition)

  • KIM, NAMHEE. A semiparametric statistical approach to Functional MRI data. 2009. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1262295445.

    MLA Style (8th edition)

  • KIM, NAMHEE. "A semiparametric statistical approach to Functional MRI data." Doctoral dissertation, Ohio State University, 2009. http://rave.ohiolink.edu/etdc/view?acc_num=osu1262295445

    Chicago Manual of Style (17th edition)