An outstanding issue in graph-theoretical studies of brain functional connectivity is the lack of formal criteria for choosing parcellation granularity and correlation threshold. Here, we propose detectability of scale-freeness as a benchmark to evaluate time-series extraction settings. Scale-freeness, i.e., power-law distribution of node connections, is a fundamental topological property that is highly conserved across biological networks, and as such needs to be manifest within plausible reconstructions of brain connectivity. We demonstrate that scale-free network topology only emerges when adequately fine cortical parcellations are adopted alongside an appropriate correlation threshold, and provide the full design of the first open-source hardware platform to accelerate the calculation of large linear regression arrays.
Keywords: Functional connectivity; Functional magnetic resonance imaging (fMRI); Graph-based analysis; Network topology; Parallel processing; Scale freeness.
Copyright © 2013 IPEM. Published by Elsevier Ltd. All rights reserved.