NEW TECHNIQUE FOR FMRI PROCESSING

University Of Minnesota Twin Cities, Minneapolis MN

Investigators

Linked publications & trials

Abstract

DESCRIPTION (Adapted From The Applicant's Abstract): Functional magnetic resonance imaging (fMRI) is becoming a widely accepted methodology for studying brain function in both normal and diseased states. While many technical advances have been made in fMRI in the past several years, there are still technical limitations associated with the methodology. One aspect of fMRI that calls for further research is data processing. To date, the majority of the approaches for extracting signal change related to neuronal activity from MRI data rely on some prior knowledge of the fMRI response. These approaches are inadequate for application where the fMRI response is not known a priori and/or complex. Although model-free approaches have been previously introduced, further development is needed to make them robust and suitable for routine use. The goal of this project is to explore a new clustering approach based on self-organizing map (SOM) algorithms for identifying and detecting neural activity related responses in fMRI data. Specifically, we will 1) incorporate pixel connectivity in the SOM method, 2) develop statistical method(s) for optimally merging nodes in the SOM and testing the significance of the resultant clusters using approaches analogous to analysis of variance, and 3) validate the proposed approach with experimental data from paradigms that call for model free analysis methods. the success of this project will substantially expand the applications of fMRI by providing the means for detecting novel responses and permitting the use of flexible paradigm designs.