Functional network stability and average minimal distance – A framework to rapidly assess dynamics of functional network representations

Publication date: 15 February 2018
Source:Journal of Neuroscience Methods, Volume 296
Author(s): Jiaxing Wu, Quinton M. Skilling, Daniel Maruyama, Chenguang Li, Nicolette Ognjanovski, Sara Aton, Michal Zochowski
BackgroundRecent advances in neurophysiological recording techniques have increased both the spatial and temporal resolution of data. New methodologies are required that can handle large data sets in an efficient manner as well as to make quantifiable, and realistic, predictions about the global modality of the brain from under-sampled recordings.New methodTo rectify both problems, we first propose an analytical modification to an existing functional connectivity algorithm, Average Minimal Distance (AMD), to rapidly capture functional network connectivity. We then complement this algorithm by introducing Functional Network Stability (FuNS), a metric that can be used to quickly assess the global network dynamic changes over time, without being constrained by the activities of a specific set of neurons.ResultsWe systematically test the performance of AMD and FuNS (1) on artificial spiking data with different statistical characteristics, (2) from spiking data generated using a neural network model, and (3) using in vivo data recorded from mouse hippocampus during fear learning. Our results show that AMD and FuNS are able to monitor the change in network dynamics during memory consolidation.Comparison with other methodsAMD outperforms traditional bootstrapping and cross-correlation (CC) methods in both significance and computation time. Simultaneously, FuNS provides a reliable way to establish a link between local structural network changes, global dynamics of network-wide representations activity, and behavior.ConclusionsThe AMD-FuNS framework should be universally useful in linking long time-scale, global network dynamics and cognitive behavior.



http://ift.tt/2lPybMK