Improving the Sensitivity of Task-Related Functional Magnetic Resonance Imaging Data Using Generalized Canonical
Emmanouela Kosteletou1, Panagiotis G Simos2,3, Eleftherios Kavroulakis4
1Institute of Applied and Computational Mathematics, Foundation for Research and Technology - Hellas (FORTH), Heraklion, Greece.
Frontiers in Human Neuroscience
|December 31, 2021
Summary
Generalized Canonical Correlation Analysis (gCCA) enhances functional Magnetic Resonance Imaging (fMRI) analysis sensitivity compared to General Linear Modeling (GLM). This method reveals more activation in brain regions during action observation tasks.
Area of Science:
- Neuroimaging
- Cognitive Neuroscience
- Biostatistics
Background:
- General Linear Modeling (GLM) is standard for fMRI signal detection but ignores spatial dependencies.
- Multivariate methods like Generalized Canonical Correlation Analysis (gCCA) address GLM limitations by considering spatial correlations.
Purpose of the Study:
- To evaluate the sensitivity improvement of GLM by applying gCCA to fMRI data.
- To compare gCCA with standard GLM in detecting brain activation during action observation tasks.
Main Methods:
- Applied gCCA to preprocessed fMRI data from a block-design experiment.
- Analyzed data from 25 healthy volunteers performing two action observation tasks at 1.5T.
- Conducted whole-brain and subject-level ROI analyses.
Main Results:
- gCCA demonstrated significantly higher activation intensity in multiple brain regions for both tasks.
- gCCA revealed activation in the primary somatosensory and ventral premotor areas during action observation.
- Subject-level ROI analyses showed gCCA improved signal-to-noise ratio and activation extent.
Conclusions:
- gCCA offers improved sensitivity over conventional GLM for task-based fMRI.
- gCCA is a promising multivariate technique for enhancing fMRI data analysis.
- This method aids in identifying subtle or spatially dependent activation patterns.
