Analysis of functional magnetic resonance imaging data using signal processing techniques

Sea Chen, Purdue University

Abstract

The goal of this research was to develop a set of tools for the analysis of functional magnetic resonance imaging data. The study of the blood oxygenation level dependent (BOLD) signal response was specifically targeted. In the first part of this research, we developed an amplitude independent clustering strategy called clustered components analysis. This technique accounted for activation amplitude variations due to partial volume effects and magnetic field inhomogeneities. The analysis framework also included an automated method for determining the number of clusters based upon the minimum description length criterion. The technique was implemented using the expectation-maximization algorithm. In the second part of this research, we, introduced an analysis method to obtain a high temporal resolution estimate without using a short time-to-repetition (TR). This method that we call supertemporal resolution analysis was developed to reduce the distortion of the BOLD response by blood inflow effects. The technique was based upon maximum a posteriori (MAP) estimation utilizing Bayesian prior model to implement temporal regularization. A crossvalidation strategy was used to automatically determine from the data the level of smoothing. A novel data simulator was developed to test our methods.

Degree

Ph.D.

Advisors

Lowe, Purdue University.

Subject Area

Biomedical research|Electrical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS