Tor Dessart Wager

This paper shows that modular fMRI preprocessing pipelines can reintroduce artifacts by sequentially projecting data into non-orthogonal subspaces, and suggests remedies like combining steps into a single filter or orthogonalizing them.

Subsampling 491 Human Connectome Project subjects puts numbers on fMRI group analysis: roughly 40 subjects catch large effects (d > 0.8), near 80 are needed for medium ones.

This study performs a large-scale meta-analysis of fMRI data to map functional-anatomical regions in the lateral frontal cortex, finding low regional modularity and suggesting distributed network processing.

Meta-analysis across human neuroimaging studies in four task domains shows preferential activation of anatomically defined ventral anterior, dorsal anterior, and posterior insula sub-regions across emotion, pain, and cognitive control.