I have ERP data, with about 80 trials, each trial epoched around an event -1 to 1 seconds.
I am interested in assessing how the connectivity changes over a course of a treatment. Accordingly, I would like to use AEC as my measure of task-based functional connectivity. However, my question is because I have a task and not resting state eeg, would it be wrong for me to use the option 'Concatenate input files before processing'? Should I instead be using the first ('save individual results') or third ('save average connectivity matrix') option?
I do see this on the Connectivity webpage which makes me think I'm wrong and that I should be concatenating:
Output configuration: Generally, the above calculation results in a 4-D matrix, where dimensions represent channels (1st and 2nd dimensions), time points (3rd dimension), and frequency (4th dimension). In the case that we analyze event-related data, we have also several files (trials). However, due to poor signal to noise ratio of a single trial, an individual realization of connectivity matrices for each of them is not in our interests. Consequently, we need to average connectivity matrices among all trials of a specific event. The second option of this part performs this averaging.
Any provided insight would be greatly appreciated.
Indeed, 2s epochs are typically too short for meaningful outcomes from most connectivity measures. If you are looking at connectivity changes post-treatment, then I would recommend you compute connectivity from longer time segments, possibly entire runs.
Due to the design and epoching of the ERP data, unfortunately I cannot extend it any further. Would concatenating be okay then? Or should I not be assessing connectivity at all with my data?
sorry to revive an old thread, but I was reading the connectivity tutorial page again, and noticed an update of information which I felt was relevant to assessing task-based connectivity.
Under the "TODO : Connectivity measure on real data : MEG/EEG data" section, it mentions assessing the connectivity of the time 2 window (100 - 300 ms), and in the shared process image, it shows the time window being 0 - 100 ms (which I'm assuming is meant for time 1 window).
I have a TMS-EEG dataset in which I want to investigate the following: does stimulation of the (left) inferior parietal lobule lead to a functional connectivity pattern that anatomically resembles the typical default mode network (and if so, in what frequency band).
Based on the previous literature, it looks like amplitude envelope correlation (AEC) would be a suitable metric to capture large-scale resting state networks with M/EEG data, with an ROI-to-vertice approach (1 x N).
My questions (and apologizes if this is beyond the scope of this forum) is then:
My data is currently epoched with 80 trials from - 1 s to 1 s for each trial. TMS-evoked potentials typically lasts up to 500 ms at the latest. Since I cannot find any literature to suggest a suitable time window of analysis in a similar context, I was thinking of using a time window of 20 ms (not 0 to avoid TMS-related artifacts) to 500 ms. I'm assuming this is too small of a time window. My question is then would concatenating be okay in this situation to study the connectivity of my regions of interest? Or even then, is my resulting length inadequate? Would it be better to use the entire post-TMS epoch (i.e. 20 - 1000 ms)?
I think concatenation would be a correct approach indeed. The data length would be therefore much greater and you would be better equipped to detect connectivity effects with AEC even on the lower end of the frequency spectrum.