fMRI GLM Analysis Guide

To specify a design matrix for fMRI GLM analysis, you must select an appropriate HRF basis set, apply high-pass filtering to preserve task frequencies, and define valid contrasts. This guide provides a domain-validated decision tree to vet these first-level pipeline choices.

The best way to handle confound regression in task-based fMRI is to include motion regressors, spike modeling, and CompCor components. This guide outlines protocols for these confounds alongside valid prewhitening strategies for modern TRs.

You need derivative or FIR basis sets for hemodynamic response modeling only when canonical HRF assumptions are violated. This guide ensures these advanced basis functions are chosen only when empirically justified to maintain statistical integrity.

Choosing between cluster-based, TFCE, and permutation correction for statistical inference depends on your spatial assumptions and distribution normality. This guide clarifies how to apply these multiple comparison correction criteria across first and second-level models.

Prewhitening matters for autocorrelation in fMRI GLM because unmodeled temporal dependencies inflate statistical significance. This guide provides valid prewhitening strategies to address autocorrelation and preserve the integrity of your statistical inference.

Yes, this fMRI analysis guide supports COBIDAS reporting by providing a checklist to document software versions, smoothing kernels, contrast weights, and statistical thresholds, ensuring your research outputs meet neuroimaging reporting standards.