Related Experiment Video
Updated: Oct 6, 2025
08:45
Mapping Cortical Dynamics Using Simultaneous MEG/EEG and Anatomically-constrained Minimum-norm Estimates: an Auditory Attention Example
Published on: October 24, 2012
14.8K
Spatial Bayesian GLM on the cortical surface produces reliable task activations in individuals and groups
Daniel Spencer1, Yu Ryan Yue2, David Bolin3
1Department of Statistics, Indiana University, Myles Brand Hall E104 901 E. 10th Street Bloomington, IN, 47408, USA.
Neuroimage
|January 15, 2022
Summary
The spatial Bayesian GLM improves brain activation detection by using neighboring data, offering reliable results for individual and group analyses, even with small sample sizes.
Area of Science:
- Neuroimaging
- Computational Neuroscience
- Statistical Modeling
Background:
- General Linear Model (GLM) is standard for brain activation analysis but is univariate and can be underpowered.
- Classical GLM struggles with noisy estimates and detecting subtle activations, especially in individuals.
Purpose of the Study:
- Assess reliability and power of a surface-based spatial Bayesian GLM for task activation.
- Compare its performance against traditional methods, particularly for individual and group analyses.
Main Methods:
- Applied a cortical surface-based spatial Bayesian GLM to motor task fMRI data from 45 Human Connectome Project subjects.
- Utilized subject-specific cortical surfaces and extended the model for multi-run analysis.
Main Results:
- Spatial Bayesian GLM yielded highly reliable individual subject activations and detected trait-like functional topologies.
- Enhanced group-level analysis reliability in samples as small as 45 subjects.
- Activation detection power remained high and nearly invariant to sample size, even for n=10.
Conclusions:
- Surface-based spatial Bayesian GLM offers superior reliability and power for brain activation analysis compared to classical GLM.
- The method is computationally efficient, reliable across sample sizes, and accessible via the BayesfMRI R package.
