Connectome of autistic brains, global versus local characterization

Abstract

The underlying neural mechanisms of autism spectrum disorders (ASD) remains unclear. Most of the previous studies based on connectomics to discriminate ASD from typically developing (TD) subjects focused either on global graph metrics or specific discriminant connections. In this paper we investigate whether there is a correlation between local and global features, and whether the characterization that discriminates ASD from TD subjects is primarily given by widespread network differences, or the difference lies in specific local connections which are just captured by global metrics. Namely, whether miswiring of brain connections related to ASD is localized or diffuse. The presented results suggest that the widespread hypothesis is more likely.