Chunk #31 — Connectivity methods — Proposed analysis algorithms to investigate directional connectivity in layer-dependent connectivity data — Iterative ICA (in relation to Fig. 8).
Functional brain networks usually incorporate multiple local and distant brain areas at a macroscopic spatial scale. It has been shown, however, that functional networks can be further separated into smaller sub-networks (Braga and Buckner 2017; Heinzle et al. 2011; Smith et al. 2009). With CBV-based, submillimeter fMRI data, it becomes possible to investigate the topographical sub-division patterns of larger functional networks into smaller and smaller units, without unwanted signal leakage from macro veins. Here, we used iterative FSL Melodic (Multivariate Exploratory Linear Optimized Decomposition into Independent Components) ICA decomposition (Beckmann and Smith 2004) to extract functional networks across multiple macroscopic and mesoscopic spatial scales. We focused on individual manually selected components and iteratively decompose them into smaller and smaller sub-components.
