My protocol consist in a motor task cued by a bip cue, in healthy and clinical population.
So I have 3 times of interest the "BIP’, the onset and the offset of the movement.
So I have different epoch length because across all the trials and all the subjects the time to achieve the motor task is different.
For my analysis, I plan to use ERD/ERS with TF by morlet wavelet.
The first question is : is it right to do TF maps across trials with same epoch length but event of interest not in the same time ? Is there a cancellation phenom ?
The second question is : is it right to import different epoch length (how to in BST ? ), them interpolate or upsample (with antiliasiang) to the longest trials, to match the onset or the offset of the movement across all the trials ? And them use morlet for TF maps ?
Thanks for your interests, and for sharing idea / comments.
Importing epochs of different length in Brainstorm is not straightforward. You need to create manually extended events first, then import them all. If you have epochs of different length in the database, select them all in Process1 and use process Standardize > Uniform epoch time.
However, you really need to question what you are expecting to get in output of this processing pipeline. If you reinterpolate an epoch on a different duration, you will heavily distort the frequency spectrum. Your various events might align better but the frequencies won’t anymore. It also means you artificially extend or shrink the duration of various brain processes.
I think that in a standard ERP setup like yours, it is maybe better to focus only on one thing at a time: either you study primarily the perceptual part (analysis time-locked on the stim trigger), or the motor part (time-locked on the response trigger). In both cases, import epochs that are long enough to include all the other events across all the trials, then compute the average of the TF decomposition, as explained in the tutorials. The other events won’t be aligned across the epochs, but if they correspond to oscillations at a specific frequency, averaging in power would still enhance them. https://neuroimage.usc.edu/brainstorm/Tutorials/TimeFrequency#MEG_recordings:_Single_trials