I´m trying to find a global measure of similarity for spikes trains over time. The signals look like in the picture (in this example I have 17 neurons).
Can I use windowed cross correlations? If yes, what should I do with the output matrices? I use MATLAB by the way.
Yes, you can use binned spikes or binned bursts to compute the cross correlations. You can then threshold the output matrix to see what electrodes have the most similarity.
If you have the spikes in text format or have exported your spike train to an HDF5 file from MC_Rack, you can use MEAnalyzer to compute and export all the comparisons: https://github.com/RDastgh1/MEAnalyzer (but please cite it if you use it).
