- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Theses and Dissertations /
- Novel region based fMRI analysis using invariant moment...
Open Collections
UBC Theses and Dissertations
UBC Theses and Dissertations
Novel region based fMRI analysis using invariant moment descriptors Ng, Bernard
Abstract
A new functional magnetic resonance imaging (fMRI) analysis approach based on invariant spatial descriptors was developed for quantitative characterization of brain activation patterns within a region of interest (ROI). In particular, the feasibility of using three dimensional moment invariants (3DMIs) to perform such spatial characterization was examined. The use of spatial descriptors is particularly novel in the field of ROI-based fMRI analysis, since up till now, only magnitude-based features were traditionally employed, which neglect the information encoded by voxel locations within an ROI. The invariance properties of the proposed descriptors to similarity transformations account for inter-subject variability in brain size and subject's orientation within the MR scanner, thus allowing for spatial distributions of activation statistics to be meaningfully compared across subjects. Enhanced sensitivity in detecting task-related activation differences as compared to traditional magnitude-based methods was demonstrated with real fMRI data. To handle the issue of feature selection, a modified linear discriminant analysis (LDA) procedure that incorporates leave-one-out cross-validation was developed. Also, methods to deal with the two main issues in ROI-based fMRI group analysis, namely errors in ROI delineation and inclusions of voxels falsely deemed active, were proposed. One method involves remapping the coordinate space with a Gaussian function, which in effect de-emphasizes voxels near the ROI boundary, thus also accounts for inter-subject variability in brain shapes. The other method detects outlier voxels that exhibit disproportional influence on the proposed invariant spatial descriptors, and deweights or removes those voxels accordingly. Testing these processing methods on real fMRI data showed further increase in discriminability of task-related activation differences compared to the original 3DMIs alone. To fully exploit the spatio-temporal structure inherent in fMRI data, we extended our spatial characterization approach into the spatio-temporal domain. We showed, for the very first time, that the modulation of the spatial distribution of BOLD signals does in fact correlate with the stimulus, and provides greater sensitivity in detecting activated ROIs and in discriminating task-related differences as compared to traditional mean intensity-based methods. The neuroscience implications of our findings are substantial, and might hence provide brain researchers and clinicians a new promising direction to explore.
Item Metadata
Title |
Novel region based fMRI analysis using invariant moment descriptors
|
Creator | |
Publisher |
University of British Columbia
|
Date Issued |
2007
|
Description |
A new functional magnetic resonance imaging (fMRI) analysis approach based on invariant spatial descriptors was developed for quantitative characterization of brain activation patterns within a region of interest (ROI). In particular, the feasibility of using three dimensional moment invariants (3DMIs) to perform such spatial characterization was examined. The use of spatial descriptors is particularly novel in the field of ROI-based fMRI analysis, since up till now, only magnitude-based features were traditionally employed, which neglect the information encoded by voxel locations within an ROI. The invariance properties of the proposed descriptors to similarity transformations account for inter-subject variability in brain size and subject's orientation within the MR scanner, thus allowing for spatial distributions of activation statistics to be meaningfully compared across subjects. Enhanced sensitivity in detecting task-related activation differences as compared to traditional magnitude-based methods was demonstrated with real fMRI data. To handle the issue of feature selection, a modified linear discriminant analysis (LDA) procedure that incorporates leave-one-out cross-validation was developed. Also, methods to deal with the two main issues in ROI-based fMRI group analysis, namely errors in ROI delineation and inclusions of voxels falsely deemed active, were proposed. One method involves remapping the coordinate space with a Gaussian function, which in effect de-emphasizes voxels near the ROI boundary, thus also accounts for inter-subject variability in brain shapes. The other method detects outlier voxels that exhibit disproportional influence on the proposed invariant spatial descriptors, and deweights or removes those voxels accordingly. Testing these processing methods on real fMRI data showed further increase in discriminability of task-related activation differences compared to the original 3DMIs alone. To fully exploit the spatio-temporal structure inherent in fMRI data, we extended our spatial characterization approach into the spatio-temporal domain. We showed, for the very first time, that the modulation of the spatial distribution of BOLD signals does in fact correlate with the stimulus, and provides greater sensitivity in detecting activated ROIs and in discriminating task-related differences as compared to traditional mean intensity-based methods. The neuroscience implications of our findings are substantial, and might hence provide brain researchers and clinicians a new promising direction to explore.
|
Genre | |
Type | |
Language |
eng
|
Date Available |
2011-02-25
|
Provider |
Vancouver : University of British Columbia Library
|
Rights |
For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.
|
DOI |
10.14288/1.0100697
|
URI | |
Degree | |
Program | |
Affiliation | |
Degree Grantor |
University of British Columbia
|
Campus | |
Scholarly Level |
Graduate
|
Aggregated Source Repository |
DSpace
|
Item Media
Item Citations and Data
Rights
For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.