Valid inference for group analysis of functionally aligned fMRI images

Functional magnetic resonance imaging (fMRI) data require preprocessing steps before statistical analysis. Multi-subjects fMRI studies are complicated: the brain’s anatomical and functional structure varies across subjects. Anatomical alignment does not capture the functional variability across subjects; the functional alignment is then applied. Generally, group analysis on functionally aligned fMRI data refers to between-subject classification. We propose an inference group analysis arguing that using functional aligned images based on Procrustes transformation does not affect type I error.