fMRI
Topics
Reference
Unix for Neuroimagers
FSL Tutorial
Principles of fMRI
Methods
Notes
Distinct analytic approaches Girn2023
- ICA
- seed-based FC
- Interregional FC
- hard parcellations with binary network assignments (typically from Yeo2011)
- Leading eigenvector dynamics analysis → LEiDA
ICA, seed-based FC, and LEiDA → soft parcellations (generate weighted network assignments) - ICA often uses dual regression with group-level network maps derived from Smith2009
- seed-based FC and LEiDA do not rely on group-level network templates
Brain exhibits significant transient states that are obscured by session-level averaging Lurie2018
Differences in statistical assumptions and conception of network definition renders comparisons of results across techniques nontrivial Girn2023
Use of a priori templates & assignments from subjects during sober states may limit network estimation in altered states Girn2023
Network approaches to neuroimaging analysis don’t often acknowledge individual differences → we have an expectation that regions are basically the same across people Girn2023 Eickhoff2018
Univariate & bivariate network approaches don’t account for multivariate dependencies
- Whole-brain spatiotemporal average FC of a given region or voxel may be more appropriate → global fC
- Preller2020, Tagliazucchi2016, Girn2022 applies this to LSD
- Preller2020, Girn2022, Madsen2021 applies this to psilocybin
- Timmermann2023 applies this to DMT