[Eeglablist] Single Trial Time Variant Connectivity

[Michael Glassen]

Hi List,

I have an analysis problem I've run into that I want to see if anyone can help with. I am working with data from a 1 subject study with very specific interventions that we have added EEG onto. We are only able to record for a very limited amount of time each session, and during that time the subject performs three trials of a movement sequence that last about 25 seconds. All of the past EEG studies I have worked on have been more traditionally designed for EEG, with many trials where time frequency/connectivity outcome measures can be averaged together to help tease out any effects. This is not possible with this data, as there are no repeated movements in the sequence performed, so the maximum trials I would have is three to average(after converting to frequency domain and time warping). I have calculated some time frequency graphs for this data and we are able to see some effects that look linked to the movements(understandably with noise attached). I also tried using a moving window to calculate time variant coherence, but the results were too random to attempt to draw any conclusions from. Realizing this is not the ideal dataset, does anyone know of any connectivity measures or techniques that have enough stability to glean info from a single/very few trials?

Best,
Mike