mdhumphries/SpikeTr​ainCommunitiesToolB​ox

MATLAB toolbox for spike-train community detection

A set of functions for analysing large-scale recordings of cellular-level neural activity, based on community detection ideas from network theory. The current version is based on Bruno, Frost & Humphries (2015, Neuron). See the Documentation folder for full details.
Motivation:

Large-scale recording technology for single-cell activity is now routinely available in variety of methods: silicon probes, multi-electrode arrays, tetrodes, calcium imaging, and voltage-sensitive dye imaging. Having captured the activity of large populations of neurons at single-cell resolution, the next question is: how do I analyse that data?

Key to that analysis is dimension-reduction. One approach to dimension-reduction is to use the fact that neurons tend to fire together in groups - or "ensembles".

We showed how the idea of community detection on arbitrary networks are ideally suited to solve the problem of detecting neural ensembles (Humphries, 2011).

The purpose of this toolbox is to develop the community-detection algorithms best-suited for the ensemble-detection problem.

The original code released with Humphries (2011) has been updated to include "consensus" community detection, that dramaticaly improves the reiability of the clustering.