Speaker: Aaron Hudson, Graduate Student, UW Biostatistics
Abstract: Advancements in neuroimaging technology have stimulated investigation into functional brain connectivity – the temporal dependence of neurological activity within brain regions which may be physically separated. By measuring neuronal activity at many brain regions over time, we can estimate the functional connectivity within a single brain. The resulting data takes the form of a network for each subject within a sample; nodes represent regions, and edges represent functional connectivity.
It is desirable to define one single network which describes the central tendency – or “average” – of a population of functional connectivity networks. Such averages can be used to examine variability in the typical network across populations.
In this presentation, I review an approach proposed by Ginestet et. al (2017) to define the average network and perform hypothesis tests for differences in the average. The authors carefully characterize the space networks and use the Frechet mean to define a notion of average on this space. Utilizing asymptotic theory for Frechet means, the authors develop a fully nonparametric method for estimation and testing. The proposed methodology is used to analyze funtional neuroimaging data from the 1000 Functional Connectomes Project.