Back to search

University of British Columbia

Constrained principal component analysis for detection of task-based cognitive modes with functional MRI : a comparative analysis of traditional and novel methods

Abstract

dc:description

Functional magnetic resonance imaging (fMRI) is a powerful neuroimaging technique for studying the spatial and temporal aspects of blood-oxygen level dependent (BOLD) signal changes during performance of specifically designed tasks. To date, task-based fMRI studies have reliably identified twelve cognitive modes, which are anatomical configurations with associated functions that underlie the cognitive, sensory, and motor processes during task performance. To detect these modes, our research group has been employing a traditional method called G-matrix Constrained Principal Component Analysis for fMRI (fMRI–CPCA), which constrains the variance in BOLD signal to variance related to task timing. However, problems can arise if different modes get merged, or when a mode is poorly formed due to the substantially small variance it accounts for. By relying on the now well-established anatomical details of the twelve cognitive modes, we have developed a novel detection method called H-matrix fMRI–CPCA, which constrains the variance in BOLD signal spatially to the prototypical templates of twelve modes. To assess this new method in comparison with G-matrix fMRI–CPCA, we used both methods to analyze fMRI data for healthy adults’ performance of five selected tasks: Antonyms, Letter Comparison, Pattern Comparison, Digit Symbol, and Sensorimotor. G-matrix method produced components in which the modes were merged when they typically get separated into different components. H-matrix method redistributed these modes over more components but may have also produced Type I errors. For Antonyms, Letter Comparison, Pattern Comparison, and Digit Symbol, some merging of modes remained, and interpretation was not improved. Sensorimotor task showed better temporal and spatial separation of modes, and interpretation improved in H-matrix analysis, which was attributable to task design. Antonyms, Letter Comparison, Pattern Comparison, and Digit Symbol were designed similarly and hence shared common weaknesses, specifically the short inter-stimulus interval (0.5s or 0.25s), whereas Sensorimotor used longer intervals (2-26s jittered). Since the selected sample had a wide age range (about 20-80), age was investigated in a secondary analysis. Results from G-matrix analysis suggested that Language, Response, and Default Modes A and B showed age differences, potentially reflecting changes in vocabulary skills and cognitive processing speed associated with healthy ageing.

Degree

thesis:*
Name thesis:degree_name
Master of Science - MSc
Level thesis:degree_level
master's
Discipline thesis:degree_discipline
Neuroscience
Grantor dc:publisher
University of British Columbia
Year dc:date
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Ly, Huan Nhat Nguyen
Contributors dc:contributor
  • Woodward, Todd S.

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial-NoDerivatives 4.0 International
Language dc:language
eng

Identifiers

dc:identifier.*
Handle dc:identifier
http://hdl.handle.net/2429/90753

Chain of custody

source
Harvested from
University of British Columbia
Base URL
circle.library.ubc.ca/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Ly, Huan Nhat Nguyen. Constrained principal component analysis for detection of task-based cognitive modes with functional MRI : a comparative analysis of traditional and novel methods. master's thesis, University of British Columbia, 2025. http://hdl.handle.net/2429/90753